Standing Records
Every record behind every Standing on the profiles: the act, the step it reaches on its ladder, and the document that proves it. Search by name, by kind of act, or by how high it reaches.
A profile’s Standing is the highest step its records reach on a fixed ladder, component by component. These are the records themselves, all of them: what the subject can make others do, how hard the seat is to remove, and what of theirs keeps running. A press or scholarly source counts one step lower than a filing, a statute or the subject’s own page; an ended seat counts only toward durability. Search a name to see a whole file, or a kind of act (“voting control”, “statutory”, “contract”) to see who holds it. The ladders ranked by these records are on the profiles index.
| File | Component | Step | Record | Source |
|---|---|---|---|---|
| 10A LABS | entrenchment | 6 | operating — 10a Labs' own release says the company, founded September 2022 and headquartered in Washington, DC, serves frontier AI labs and other customers and will open offices in London, Tokyo and Sydney in Q1 2027. | first-hand 2026-09-16 |
| 10A LABS | reach | 6 | output acted on — 10a Labs evaluator-for Anthropic | official |
| 23ANDME | entrenchment | 6 | operating — An 8-K states the 23andMe Personal Genome Service and Research Services businesses were sold to TTAM Research Institute (now 23andMe Research Institute) for $305.0 million in cash, closing 14 July 2025. | official 2025-07-14 |
| 23ANDME | entrenchment | 6 (ended) | operating — The Form 10-K for the year ended 31 March 2025 reports about 14.0 million customers and dual-class stock (20,499,552 Class A and 7,041,942 Class B shares outstanding) for 23andMe Holding Co., which filed Chapter 11 on 23 March 2025. | official 2025-03-31 |
| AADHAAR / UIDAI / INDIA STACK | entrenchment | 10 | statute or infrastructure — UIDAI is a statutory authority established by the Aadhaar Act, and the Supreme Court's 26 September 2018 judgment upheld the Act in part while striking down Section 57 and Section 33(2), after which Parliament amended the Act in 2019 rather than repealing it. | official 2018-09-26 |
| AADHAAR / UIDAI / INDIA STACK | funding insulation | 8 | endowment or appropriation — UIDAI's 2024-25 annual report says it receives Grants-in-Aid from MeitY (600.00 crore rupees in 2024-25) and that a separate UIDAI fund has been constituted under Section 25 of the Act for its financial autonomy. | official 2025-03-31 |
| AADHAAR / UIDAI / INDIA STACK | reach | 10 | statutory power — The Aadhaar and Other Laws (Amendment) Act, 2019 (Act 14 of 2019), amending the Aadhaar Act 2016, writes the authentication regime, including new Section 4(3)-(7), into statute for the Unique Identification Authority of India. | official 2019-07-23 |
| ADAM GLEAVE | durability | 6 | built operating — FAR.AI, the AI safety research institute Gleave co-founded, operates and lists him as Co-founder & CEO. | first-hand 2022-01-01 |
| ADAM GLEAVE | durability | 3 | cited work — Co-first author of 'Adversarial Policies Beat Superhuman Go AIs' (ICML 2023), in which an adversary won over 97 percent of games against KataGo. | scholarly 2023-07-23 |
| ADAM GLEAVE | insulation | 4 | appointed officer — FAR.AI's team page lists Adam Gleave as Co-founder & CEO, under both Leadership and Board. | first-hand |
| ADAM GLEAVE | insulation | 2 | at will employee — Adam Gleave employed-by DeepMind | official |
| ADAM GLEAVE | leverage | 6 | runs org acted on — FAR.AI's team page lists Adam Gleave as Co-founder & CEO and on its board; OpenAI's GPT-5 system card reports FAR.AI red-teamed the OpenAI API for 80 hours and OpenAI states 'We are actioning on some of these system vulnerabilities'. | first-hand 2025-08-07 |
| ADAM GLEAVE | leverage | 4 | advisory seat — Adam Gleave advisory European AI Office | official |
| ANTI-DEFAMATION LEAGUE (ADL) | entrenchment | 6 | operating — ADL's Form 990 for the year ended 30 June 2025 reports total revenue of $135,300,377, net assets of $50,454,946 and 623 employees. | official 2025-06-30 |
| ANTI-DEFAMATION LEAGUE (ADL) | funding insulation | 6 | independent funding — ADL's Form 990 for the year ended 30 June 2025 reports $135,485,438 in contributions and grants, a blank government-grants line, program-service revenue of $427,368 and a 21-member voting board of whom 20 are independent. | official 2025-06-30 |
| ANTI-DEFAMATION LEAGUE (ADL) | reach | 6 | output acted on — YouTube's statement to the Senate Commerce Committee says it expanded its Trusted Flagger Program to an additional 50 NGOs in 2017, including the Anti-Defamation League, whose members are given a bulk-flagging tool to alert YouTube to content that may violate its policies. | official 2018-01-17 |
| ANTI-DEFAMATION LEAGUE (ADL) | reach | 6 | output acted on — Anti-Defamation League contract US Department of Homeland Security | official |
| AGUSTÍN CARSTENS | durability | 2 | publications — Carstens co-authored BIS Working Paper 1178, Finternet: the financial system for the future, with Nandan Nilekani. | official 2024-04 |
| AGUSTÍN CARSTENS | insulation | 4 (ended) | appointed officer — The BIS Board of Directors appointed his successor as General Manager and announced that Carstens's term concludes at the end of June 2025. | official 2024-11-11 |
| AGUSTÍN CARSTENS | leverage | 6 (ended) | runs org acted on — The BIS lists Carstens as General Manager of the Bank for International Settlements from 1 December 2017 to 30 June 2025. | official 2025-06-30 |
| AGUSTÍN CARSTENS | leverage | 4 | advisory seat — GFTN's leadership page lists Carstens on its International Advisory Board. | first-hand |
| AI EVALUATOR FORUM | entrenchment | 6 | operating — The Forum's website carries the AEF-1 standard and a September 2026 letter urging frontier AI companies to embed third-party evaluators. | first-hand 2026-09-01 |
| AI EVALUATOR FORUM | funding insulation | 6 | independent funding — The Forum's support page says it has operated without funding since launch, sustained entirely by the voluntary participation of its members. | first-hand |
| AI EVALUATOR FORUM | reach | 4 | convenes or funds — The AI Evaluator Forum describes itself as a network of eight independent AI evaluation organizations that published the AEF-1 minimum operating conditions standard (Version 1, 4 December 2025). | first-hand 2025-12-04 |
| ARTIFICIAL INTELLIGENCE UNDERWRITING COMPANY (AIUC) | entrenchment | 6 | operating — AIUC's own team page and Series A post show an operating company that maintains AIUC-1 and conducts the technical evaluations behind certification. | first-hand 2026-09-15 |
| ARTIFICIAL INTELLIGENCE UNDERWRITING COMPANY (AIUC) | funding insulation | 6 | independent funding — AIUC says it has raised $55 million: a $15 million seed from NFDG and a $40 million Series A led by Ribbit Capital with First Harmonic and Terrain. | first-hand 2026-09-15 |
| ARTIFICIAL INTELLIGENCE UNDERWRITING COMPANY (AIUC) | reach | 6 | output acted on — AIUC's Series A post lists Cursor, ElevenLabs, Harvey, KPMG, Lovable, UiPath and Fin as companies certified to the AIUC-1 standard it maintains. | first-hand 2026-09-15 |
| ARTIFICIAL INTELLIGENCE UNDERWRITING COMPANY (AIUC) | reach | 4 | convenes or funds — AI Verification and Evaluation Research Institute (AVERI) funded-by Artificial Intelligence Underwriting Company (AIUC) | official |
| AJEYA COTRA | durability | 4 | cited work — Her 2020 draft report on AI timelines (biological anchors) predicts a median of 2052; Epoch AI published its own explainer of it. | first-hand 2020-09-18 |
| AJEYA COTRA | insulation | 2 | at will employee — Ajeya Cotra employed-by Open Philanthropy | official |
| AJEYA COTRA | leverage | 8 (ended) | signs grants — Coefficient Giving's RFP page: 'This RFP text was largely drafted by Ajeya Cotra', and the RFP 'closed after funding $25 million in grants'; she previously managed its technical AI safety grantmaking. | first-hand 2024-07-26 |
| AJEYA COTRA | leverage | 4 | advisory seat — Epoch AI's team page lists Ajeya Cotra on its Board of Directors. | first-hand 2025-01-01 |
| AJEYA COTRA | leverage | 4 | advisory seat — Ajeya Cotra board-seat Redwood Research | official |
| AJEYA COTRA | leverage | 4 | advisory seat — Ajeya Cotra advisory Schmidt Sciences | official |
| ALEX KARP | durability | 8 | program of record — The Army's enterprise agreement with Palantir runs up to 10 years with a $10 billion cap on purchases by the Army and other DoD agencies, consolidating 75 prior contracts. | official 2025-07-31 |
| ALEX KARP | durability | 7 | program of record — DefenseScoop reports Deputy Secretary Feinberg's 9 March memo directs officials to make the Maven Smart System a formal program of record by the end of the fiscal year; the May 2024 award was a $480 million five-year IDIQ, raised in May 2025 toward nearly $1.3 billion through 2029. | press 2026-04-15 |
| ALEX KARP | insulation | 8 | founder control — Palantir's 2026 proxy states that Class F shares held in the Founder Voting Trust, with the founders' other shares under the Founder Voting Agreement, 'will give our Founders the ability to control up to 49.999999% of the total voting power of our capital stock' (collective for three founders; Karp is a Founder and CEO). | official 2026-04-24 |
| ALEX KARP | leverage | 6 | runs org acted on — The U.S. Army's enterprise agreement with Palantir consolidates 75 contracts (15 prime, 60 related) into one, with volume discounts over a performance period of up to 10 years and a $10 billion cap on purchases by the Army and other DoD agencies (Karp is CEO per the DEF 14A). | official 2025-07-31 |
| ALEX STAMOS | durability | 4 | cited work — Founded SIO, whose Election Integrity Partnership 'The Long Fuse' report and the LAION-5B CSAM finding are cited outputs. | first-hand 2020-01-01 |
| ALEX STAMOS | insulation | 4 | appointed officer — Stamos announced he is joining Cognition as CISO (Cognition makes the Devin coding agent). | first-hand 2026-09-29 |
| ALEX STAMOS | insulation | 1 | at will employee — Alex Stamos employed-by Cognition | press |
| ALEX STAMOS | leverage | 6 (ended) | runs org acted on — Founding director of the Stanford Internet Observatory, which ran the Election Integrity Partnership; Stanford now lists him as 'Former Director'. | first-hand 2019-01-01 |
| ALEX STAMOS | leverage | 4 | advisory seat — His Stanford Cyber Policy Center page: 'Alex also serves on the advisory board to NATO's Collective Cybersecurity Center of Excellence in Tallinn, Estonia.' | first-hand |
| ALEXANDR WANG | durability | 6 | built operating — Founded Scale AI (2016), which still operates; Scale was the US AI Safety Institute's first authorized third-party model evaluator. | first-hand 2025-02-01 |
| ALEXANDR WANG | durability | 3 | cited work — Senior author with Dan Hendrycks and Summer Yue of Humanity's Last Exam. | scholarly 2025-01-24 |
| ALEXANDR WANG | insulation | 4 | appointed officer — Scale AI's release signs 'Alexandr Wang, Scale AI Founder and Chairman of the Board' and states the Board appointed Francis deSouza CEO effective 10 Aug 2026. | first-hand 2026-07-30 |
| ALEXANDR WANG | insulation | 4 | appointed officer — As Meta's Chief AI Officer, Wang is an appointed Meta executive. | first-hand 2025-06-30 |
| ALEXANDR WANG | insulation | 1 | at will employee — Alexandr Wang employed-by Meta | press |
| ALEXANDR WANG | leverage | 6 | runs org acted on — Meta's leadership page: Wang joined in June 2025 as the company's first-ever Chief AI Officer and leads Meta Superintelligence Labs. | first-hand 2025-06-30 |
| ALEXANDRA ABBAS | durability | 6 (ended) | built operating — Inspect Evals, the repository she led the development of, is still maintained (by Generality Labs). | first-hand 2024-10-02 |
| ALEXANDRA ABBAS | durability | 6 | built operating — Alexandra Abbas founded Arcadia Impact | official |
| ALEXANDRA ABBAS | insulation | 2 | at will employee — Alexandra Abbas employed-by Meridian Labs | official |
| ALEXANDRA ABBAS | leverage | 6 (ended) | runs org acted on — Meridian's team page: she co-founded the AI Safety Engineering Taskforce in 2024 and led development of Inspect Evals, a repository of 100+ AI evaluations (prior to joining Meridian). | first-hand 2024-01-01 |
| ALEXANDRA ABBAS | leverage | 1 | public voice — Third author of 'Transect: Retaining Observability for Long-Horizon LLM Agent Evaluations' with the UK AI Security Institute's Cozmin Ududec. | scholarly 2026-10-06 |
| ALIGNMENT RESEARCH CENTER (ARC) | entrenchment | 6 | operating — ARC's Form 990 summary reports FY2024 expenses of $9,054,183 and net assets of $3,299,798, and its team page lists current staff. | official |
| ALIGNMENT RESEARCH CENTER (ARC) | funding insulation | 6 | independent funding — ProPublica's Form 990 summary shows contributions were the whole of ARC's reported revenue in each year listed ($10,043,597 in FY2023, $5,219,369 in FY2024), and ARC says all donations are unrestricted. | official |
| ALIGNMENT RESEARCH CENTER (ARC) | reach | 4 | convenes or funds — ARC announced the ARC White-Box Estimation Challenge, a contest run with AIcrowd to improve estimation algorithms for random MLPs. | first-hand |
| AMANDA ASKELL | durability | 4 | cited work — Claude's constitution, released CC0 so anyone may use it, with her as primary author; first author of the HHH 'General Language Assistant' paper. | first-hand 2026-01-22 |
| AMANDA ASKELL | durability | 1 | publications — Amanda Askell co-author Jack Clark | scholarly |
| AMANDA ASKELL | durability | 1 | publications — Amanda Askell co-author Miles Brundage | scholarly |
| AMANDA ASKELL | durability | 1 | publications — Amanda Askell co-author Logan Graham | scholarly |
| AMANDA ASKELL | insulation | 2 | at will employee — Researcher at Anthropic (finetuning and alignment); no term or protection stated. | first-hand |
| AMANDA ASKELL | leverage | 6 | runs org acted on — Anthropic's constitution page: 'Amanda Askell is the primary author and wrote the majority of the text' of the document that Anthropic says directly shapes Claude's behavior. | first-hand 2026-01-22 |
| AMAZON RING | entrenchment | 6 | operating — The FTC reported on 23 April 2024 that it was sending more than 5.6 million dollars in refunds to Ring customers under its 2023 stipulated order, which imposed a 20-year privacy and security program. | official 2024-04-23 |
| AMAZON RING | reach | 6 | output acted on — Ring's help page says Community Requests can be posted only by public safety agencies that are Neighbors Verified and Axon Evidence users, with footage that owners choose to share landing in Axon Evidence. | first-hand |
| ANDON LABS | entrenchment | 6 | operating — Y Combinator's directory lists Andon Labs, Winter 2024 batch, San Francisco, team size 11, status Active. | first-hand |
| ANDON LABS | funding insulation | 6 | independent funding — Andon Labs says it is backed by Y Combinator (Winter 2024 batch) and that its own businesses are not profitable and it has no existing revenue-generating businesses. | first-hand |
| ANDON LABS | reach | 6 | output acted on — Andon Labs evaluator-for Anthropic | official |
| ANDON LABS | reach | 6 | output acted on — Anthropic's Opus 4.8 system card carries an External testing from Andon Labs section, and Anthropic names Andon Labs as its partner in Project Vend phase two. | official 2025-12-18 |
| ANDREA VALLONE | durability | 3 | cited work — The safe-completions approach she co-authored, tied to OpenAI's GPT-5 handling of sensitive requests. | scholarly 2025-08-12 |
| ANDREA VALLONE | insulation | 1 | at will employee — Andrea Vallone employed-by Anthropic | press |
| ANDREA VALLONE | leverage | 5 (ended) | runs org acted on — Wired: she headed OpenAI's model policy safety-research team, which led its October 2025 mental-health response report. | press 2025-11-24 |
| ANDREA VALLONE | leverage | 1 | public voice — Sixth of seven authors of 'From Hard Refusals to Safe-Completions' (arXiv 2508.09224). | scholarly 2025-08-12 |
| ANDREESSEN HOROWITZ (A16Z) | entrenchment | 6 | operating — a16z Capital Management, L.L.C. operates as an SEC-registered investment adviser co-founded by Marc Andreessen and Ben Horowitz and filed Form D for its funds in 2026. | first-hand 2026-01-09 |
| ANDREESSEN HOROWITZ (A16Z) | funding insulation | 6 | independent funding — a16z announced it raised over $15 billion in new capital on 9 January 2026. | first-hand 2026-01-09 |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Arena (formerly LMArena) funded-by Andreessen Horowitz (a16z) | official |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Arena (formerly LMArena) says its seed round was led by a16z and UC Investments, and a16z is a named existing investor in its Series B. | first-hand 2025-05-27 |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — OpenAI funded-by Andreessen Horowitz (a16z) | official |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Safe Superintelligence Inc. funded-by Andreessen Horowitz (a16z) | official |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Anduril Industries funded-by Andreessen Horowitz (a16z) | official |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Braintrust funded-by Andreessen Horowitz (a16z) | official |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Leading the Future (super PAC, FEC C00916114) funded-by Andreessen Horowitz (a16z) | official |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Promptfoo funded-by Andreessen Horowitz (a16z) | official |
| ANDREESSEN HOROWITZ (A16Z) | reach | 4 | convenes or funds — Vals AI funded-by Andreessen Horowitz (a16z) | official |
| ANDREW LOWENTHAL | durability | 6 | built operating — liber-net, which he founded, operates as a 501(c)(3) per its own site. | first-hand |
| ANDREW LOWENTHAL | insulation | 4 | appointed officer — He is CEO of liber-net, a 501(c)(3) nonprofit, per his own bios. | first-hand |
| ANDREW LOWENTHAL | leverage | 4 | convenes — His Network Affects About page: 'I also helped coordinate the Westminster Declaration', which arose from a June 2023 meeting in Westminster, London. | first-hand 2023-06-30 |
| ANDY ZOU | durability | 5 | built operating — AI Frontiers' author bio of Andy Zou states 'He's a cofounder and CTO at Gray Swan AI.' | press |
| ANDY ZOU | durability | 4 | cited work — Meta's Muse Spark Safety & Preparedness Report states that GraySwan ran prompt-injection attacks before launch using its Agent Red Teaming (ART) benchmark, citing Zou et al., 2025, and lists Andy Zou among its authors. | official 2026-05-14 |
| ANDY ZOU | leverage | 2 | public voice — Zou is lead author of 'Universal and Transferable Adversarial Attacks on Aligned Language Models', which introduced the Greedy Coordinate Gradient attack and reports adversarial suffixes that transfer to closed models including ChatGPT, Bard and Claude. | first-hand 2023-07-27 |
| ANIKA COLLIER NAVAROLI | durability | 2 | publications — Written testimony to House Oversight and CJR columns remain available. | official 2023-02-08 |
| ANIKA COLLIER NAVAROLI | insulation | 4 | appointed officer — CJR bio: she is director of the Craig Newmark Center for Journalism Ethics and Security at Columbia Journalism School. | first-hand 2026-01-01 |
| ANIKA COLLIER NAVAROLI | leverage | 8 (ended) | signs enforcement — Her House Oversight testimony: 'the safety policy team - my team - acted as the final moderators' at Twitter. | official 2023-02-08 |
| ANIKA COLLIER NAVAROLI | leverage | 2 | public voice — Writes CJR's 'Ask Anika' column and testified to the House Oversight Committee on 8 Feb 2023. | first-hand 2026-10-22 |
| ANNA MAKANJU | insulation | 4 | appointed officer — OpenAI Foundation: Anna Makanju joined as Head of AI for Civil Society and Philanthropy in mid-April 2026. | first-hand 2026-03-24 |
| ANNA MAKANJU | insulation | 2 | at will employee — Anna Makanju employed-by OpenAI Foundation | official |
| ANNA MAKANJU | leverage | 2 | public voice — Quoted as OpenAI's VP of Global Affairs at the Frontier Model Forum launch: 'Advanced AI technologies have the potential to profoundly benefit society.' | first-hand 2023-07-26 |
| ANNE WOJCICKI | durability | 6 | built operating — Anne Wojcicki founded 23andMe Research Institute (formerly TTAM Research Institute) | official |
| ANNE WOJCICKI | durability | 6 | built operating — Anne Wojcicki founded 23andMe | official |
| ANNE WOJCICKI | insulation | 8 | founder control — TTAM Research Institute's closing release: 'The TTAM Research Institute was founded and is led by Anne Wojcicki.' | first-hand 2025-07-14 |
| ANNE WOJCICKI | leverage | 4 | funds others — Breakthrough Prize launch release names Anne Wojcicki among the six who launched the Breakthrough Prize in Life Sciences. | first-hand 2013-02-20 |
| ANTHROPIC | entrenchment | 8 | program of record — Microsoft post: 'Anthropic has committed to purchase $30 billion of Azure compute capacity'; Amazon 10-Q: AWS commitment expanded by more than $100.0 billion over 10 years (10-Q not re-fetched). | first-hand 2025-11-18 |
| ANTHROPIC | entrenchment | 6 | operating — Anthropic is a Public Benefit Corporation whose board is 'elected by stockholders and our Long-Term Benefit Trust'; run-rate revenue crossed $47 billion in May 2026. | first-hand 2026-05-28 |
| ANTHROPIC | funding insulation | 6 | independent funding — Series H: $65 billion 'led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital' at $965 billion post-money; $15B of it is hyperscaler money (profile). No customer-concentration figure found, so the 20 percent test is not shown. | first-hand 2026-05-28 |
| ANTHROPIC | reach | 8 | contract or gate — DoD CDAO awarded Anthropic a two-year prototype other transaction agreement with a $200 million ceiling. | first-hand 2025-07-14 |
| ANTHROPIC | reach | 8 | contract or gate — GSA OneGov agreement with Anthropic (12 Aug 2025) offering Claude to federal agencies for $1 (page fetched but the $1 text not matched; per the profile). | official 2025-08-12 |
| ANTHROPIC | reach | 4 | convenes or funds — MATS Research funded-by Anthropic | official |
| ANTHROPIC | reach | 4 | convenes or funds — SeedAI funded-by Anthropic | official |
| ANTONY BLINKEN | durability | 6 | built operating — Co-founder of WestExec Advisors, whose site lists co-founders and shows a 2026 copyright. | first-hand 2017-09-01 |
| ANTONY BLINKEN | insulation | 4 | appointed officer — Co-chair of McKinsey's Geopolitical Advisory Council, an appointed advisory seat. | first-hand |
| ANTONY BLINKEN | leverage | 10 (ended) | statutory authority — As 71st Secretary of State he headed the Department, administered 'under the supervision and direction of the Secretary of State' (22 USC 2651a). | official 2021-01-26 |
| ANTONY BLINKEN | leverage | 4 | advisory seat — His office site: he 'co-chairs, with Condoleezza Rice, McKinsey's Geopolitical Advisory Council' and 'serves on the board of the Center for American Progress'. | first-hand |
| APART RESEARCH | entrenchment | 6 | operating — ProPublica's Nonprofit Explorer lists Apart Research Inc under EIN 415214746, and its pages describe current Sprints, the Apart Fellowship and a Research Division. | official |
| APART RESEARCH | funding insulation | 6 | independent funding — Apart's 11 July 2025 fundraiser update says it received donations from more than 70 private donors along with support from Jaan Tallinn's foundation and AISTOF. | first-hand 2025-07-11 |
| APART RESEARCH | reach | 4 | convenes or funds — Apart Research hosts weekend research Sprints and a fellowship, and Coefficient Giving's grant feed records grants of $130,050 and $161,280 to Apart for hosting AI alignment hackathons. | first-hand |
| APOLLO RESEARCH | entrenchment | 6 | operating — Apollo's about page states 'We are now a Public Benefit Corporation to achieve our mission of secure AI development and deployment' and that it partnered with OpenAI to test o1 before public deployment (2024). | first-hand 2026-01-20 |
| APOLLO RESEARCH | funding insulation | 6 (ended) | independent funding — Apollo's own Manifund reply states 'We have received ~$1.5M from OpenPhil and ~$500k from SFF speculation grants' plus under $200k from other funders (2023 figures; Apollo has since converted to a PBC with a seed round and client fees). | first-hand 2023-01-01 |
| APOLLO RESEARCH | funding insulation | 4 | graded funding disclosed — Apollo's conflict-of-interest policy says it 'generally request[s] that the parties for whom we conduct work compensate us at a fair market value' and refuses outcome-contingent pay and grants or investments from organizations it evaluates outside those contracts. | first-hand 2025-11-26 |
| APOLLO RESEARCH | reach | 6 | output acted on — Apollo Research evaluator-for Meta | official |
| APOLLO RESEARCH | reach | 6 | output acted on — Meta's Muse Spark Safety & Preparedness Report states 'Independent third-party testing by Apollo Research found Muse Spark has the highest rate of evaluation awareness they have observed to date' and reports Apollo's covert-action, oversight-subversion and sabotage evaluations. | official 2026-05-14 |
| APOLLO RESEARCH | reach | 6 | output acted on — Apollo Research evaluator-for OpenAI | official |
| APOLLO RESEARCH | reach | 6 | output acted on — Apollo Research evaluator-for DeepMind | official |
| APOLLO RESEARCH | reach | 6 | output acted on — Anthropic's Claude 4 system card states 'we contracted with Apollo Research to assess an early snapshot for propensities and capabilities related to sabotage', and quotes Apollo's summary that the snapshot 'schemes and deceives at such high rates that we advise against deploying this model either internally or externally'. | official 2025-05-22 |
| APOLLO RESEARCH | reach | 6 | output acted on — Apollo Research evaluator-for Anthropic | official |
| ARATI PRABHAKAR | durability | 6 | built operating — Arati Prabhakar founded Actuate Innovation, Inc. | official |
| ARATI PRABHAKAR | durability | 4 (ended) | cited work — Framework for Nucleic Acid Synthesis Screening (April 2024) names her as chair of the NSTC; Executive Order 14292 (5 May 2025) directs OSTP to revise or replace it and funders to buy only from providers adhering to the updated framework. | official 2025-05-05 |
| ARATI PRABHAKAR | insulation | 4 (ended) | appointed officer — Senate-confirmed (22 September 2022) OSTP Director, Assistant to the President for Science and Technology; served to January 2025. | official 2022-09-22 |
| ARATI PRABHAKAR | leverage | 6 (ended) | runs org acted on — Directed the White House Office of Science and Technology Policy, October 2022 to January 2025, which issued the Blueprint for an AI Bill of Rights. | official 2022-10-04 |
| ARTIFICIAL ANALYSIS | entrenchment | 6 | operating — Artificial Analysis, Inc. operates artificialanalysis.ai under terms of use last revised 15 September 2026 and lists a team of 40 on its careers page. | first-hand 2026-09-15 |
| ARTIFICIAL ANALYSIS | entrenchment | 4 | cited standard — OpenAI cites the Artificial Analysis Intelligence Index and Coding Agent Index in launch posts for GPT-5.6 and GPT-6 Astra. | official 2026-07-09 |
| ARTIFICIAL ANALYSIS | funding insulation | 6 | independent funding — Artificial Analysis lists eight individual backers (Nat Friedman, Daniel Gross, Andrew Ng, Adam D'Angelo, Clem Delangue, Guillermo Rauch, Shawn Wang, Charlie Songhurst) and names no model developer as investor; its CEO says no one pays to be on the website. | first-hand |
| ARTIFICIAL ANALYSIS | reach | 6 | output acted on — Artificial Analysis evaluator-for Mistral AI | official |
| ARTIFICIAL ANALYSIS | reach | 6 | output acted on — OpenAI's GPT-5.6 post states that GPT-5.6 Sol sets a new state of the art at 80 on the Artificial Analysis Coding Agent Index, and its GPT-6 Astra page lists Artificial Analysis Intelligence Index v4.1.1 in its benchmark tables. | official 2026-07-09 |
| ARVIND NARAYANAN | durability | 3 | cited work — Co-author of the Holistic Agent Leaderboard paper (arXiv 2510.11977) and AI as Normal Technology, which he describes as the leading alternative to the superintelligence view. | scholarly 2025-10-13 |
| ARVIND NARAYANAN | durability | 2 | publications — Arvind Narayanan co-author AI Evaluator Forum | official |
| ARVIND NARAYANAN | durability | 2 | publications — Arvind Narayanan co-author Beth Barnes | official |
| ARVIND NARAYANAN | durability | 2 | publications — Arvind Narayanan co-author Chris Painter | official |
| ARVIND NARAYANAN | durability | 2 | publications — Arvind Narayanan co-author Miles Brundage | official |
| ARVIND NARAYANAN | durability | 2 | publications — Arvind Narayanan co-author Sayash Kapoor | official |
| ARVIND NARAYANAN | durability | 1 | publications — Arvind Narayanan co-author Percy Liang | scholarly |
| ARVIND NARAYANAN | durability | 1 | publications — Arvind Narayanan co-author Gillian Hadfield | scholarly |
| ARVIND NARAYANAN | durability | 1 | publications — Arvind Narayanan co-author Helen Toner | scholarly |
| ARVIND NARAYANAN | durability | 1 | publications — Arvind Narayanan co-author J.J. Allaire | scholarly |
| ARVIND NARAYANAN | insulation | 4 | appointed officer — Appointed director of Princeton's Center for Information Technology Policy effective July 2023; Siegel Professor of computer science. | first-hand 2023-07-01 |
| ARVIND NARAYANAN | insulation | 2 | at will employee — Arvind Narayanan employed-by Princeton Center for Information Technology Policy (CITP) | official |
| ARVIND NARAYANAN | insulation | 2 | at will employee — Arvind Narayanan employed-by Princeton | official |
| ARVIND NARAYANAN | leverage | 4 | advisory seat — Arvind Narayanan advisory Stanford CRFM | official |
| ARVIND NARAYANAN | leverage | 4 | advisory seat — Arvind Narayanan advisory Forecasting Research Institute (FRI) | official |
| ARVIND NARAYANAN | leverage | 2 | public voice — Statement to four US Senators (30 October 2023) and the Knight Institute essay AI as Normal Technology (15 April 2025). | first-hand 2023-10-30 |
| ASPEN INSTITUTE | entrenchment | 8 | program of record — USAspending records a $5,000,000 State Department award (SECAGD26CA0053) to Aspen Institute, Inc. from September 2026, following a $16,320,000 award for virtual youth exchanges running September 2021 to August 2026. | official 2026-09-01 |
| ASPEN INSTITUTE | entrenchment | 6 | operating — The Aspen Institute states it was founded in 1949, and its 2024 Form 990 counts 754 employees and 66 voting board members, 65 of them independent. | official |
| ASPEN INSTITUTE | funding insulation | 6 | independent funding — The Aspen Institute's calendar 2024 Form 990 reports total revenue of $443,841,184 and contributions and grants of $381,262,788, of which $7,168,895 is government grants. | official |
| ASPEN INSTITUTE | reach | 4 | convenes or funds — The Aspen Institute's Commission on Information Disorder, an Aspen Digital initiative funded by Craig Newmark Philanthropies, issued 15 recommendations to government, industry and civil society in November 2021. | first-hand 2021-11-15 |
| ATLANTIC COUNCIL | entrenchment | 6 | operating — The Atlantic Council dates itself to 1961 and its calendar 2024 Form 990 lists 370 employees, 36 voting board members and Frederick Kempe as President and CEO. | official |
| ATLANTIC COUNCIL | funding insulation | 6 | independent funding — The Atlantic Council's calendar 2024 Form 990 reports contributions of $69,956,417, of which $9,908,972 is government grants, and total revenue of $69,901,633. | official |
| ATLANTIC COUNCIL | reach | 6 | output acted on — Facebook announced that experts from the Atlantic Council's Digital Forensic Research Lab would work closely with its security, policy and product teams, and that it would use the Council's Digital Research Unit Monitoring Missions during elections. | official 2018-05-17 |
| ATLANTIC COUNCIL | reach | 6 | output acted on — Atlantic Council contract DoD | official |
| ATLANTIC COUNCIL | reach | 4 | convenes or funds — The Atlantic Council hosted the DFRLab's June 2021 summit at which former Wikimedia CEO Katherine Maher described a misinformation clearinghouse built through conversations with government and other platforms. | first-hand 2021-06-01 |
| AVERI | entrenchment | 6 | operating — ProPublica's Nonprofit Explorer lists the AI Verification and Evaluation Research Institute, EIN 38-4354994, Washington DC, with tax exemption issued July 2025 and no Form 990 data yet. | official 2025-07-01 |
| AVERI | funding insulation | 6 | independent funding — AVERI's About page states 'No donor makes up a majority of our funding' and that it does not accept cash donations or compensation from frontier AI companies or their executives. | first-hand |
| AVERI | reach | 2 | publishes — AVERI launched on 15 January 2026 with its flagship paper 'Frontier AI Auditing: Toward Rigorous Third-Party Assessment of Safety and Security Practices at Leading AI Companies', and has since published a legislative survey and a pilot report. | first-hand 2026-01-15 |
| BALLARD PARTNERS | entrenchment | 6 | operating — Ballard Partners is the registrant on Foreign Agents Registration Act registration No. 7070, active since 13 January 2022, and a Senate lobbying registrant with filings from 2017 to 2026. | official 2022-01-13 |
| BALLARD PARTNERS | funding insulation | 6 | independent funding — Senate Lobbying Disclosure Act filings show Ballard Partners reporting $88.12 million of lobbying income from 339 paying clients in 2025, the largest single client being $1.71 million (1.9 percent). | official 2025-12-31 |
| BALLARD PARTNERS | reach | 4 | registered influence — Ballard Partners is an active registrant under the Foreign Agents Registration Act (No. 7070) and files lobbying reports under the LDA. | official 2026-10-24 |
| BEN BUCHANAN | durability | 4 (ended) | cited work — NSM-25 on AI (24 October 2024), which he coordinated, was rescinded and replaced by NSPM-11 on 5 June 2026. | official 2026-06-05 |
| BEN BUCHANAN | durability | 2 | publications — Author of four books on cybersecurity and AI per his testimony; The Bitter Struggle due November 2026. | first-hand 2026-07-21 |
| BEN BUCHANAN | insulation | 4 (ended) | appointed officer — White House Special Advisor for AI on the NSC, 2021 to 2025, a presidential staff post. | official 2025-01-20 |
| BEN BUCHANAN | insulation | 2 | at will employee — Ben Buchanan employed-by Johns Hopkins School of Advanced International Studies (SAIS) | official |
| BEN BUCHANAN | insulation | 2 | at will employee — Ben Buchanan employed-by NSC | official |
| BEN BUCHANAN | leverage | 4 (ended) | advisory seat — By his own sworn testimony served on the National Security Council and as first White House Special Advisor for AI during the Biden Administration. | first-hand 2026-07-21 |
| BEN BUCHANAN | leverage | 4 | advisory seat — Ben Buchanan advisory Anthropic | official |
| BEN GOLDHABER | durability | 1 | publications — Co-author of Towards Guaranteed Safe AI (arXiv 2405.06624). | scholarly 2024-05-10 |
| BEN GOLDHABER | insulation | 2 | at will employee — Ben Goldhaber employed-by Open Philanthropy | official |
| BEN GOLDHABER | insulation | 2 | at will employee — Ben Goldhaber employed-by FAR.AI | official |
| BEN GOLDHABER | leverage | 4 | funds others — Program Officer, AI Governance and International Policy, Coefficient Giving since April 2026, focused on initiatives serving AI companies' safety and governance teams. | first-hand 2026-04-01 |
| BEN NIMMO | durability | 6 | built operating — Ben Nimmo founded Atlantic Council Digital Forensic Research Lab | official |
| BEN NIMMO | durability | 4 | cited work — His Breakout Scale (Brookings, September 2020) is used by OpenAI to rate influence operations in its 2026 case studies. | first-hand |
| BEN NIMMO | insulation | 2 | at will employee — Ben Nimmo employed-by Meta | official |
| BEN NIMMO | insulation | 2 | at will employee — Employee at OpenAI (Principal Investigator); no stated term. | first-hand 2026-06-25 |
| BEN NIMMO | insulation | 2 | at will employee — Ben Nimmo employed-by Graphika | official |
| BEN NIMMO | insulation | 2 | at will employee — Ben Nimmo employed-by NATO | official |
| BEN NIMMO | leverage | 6 (ended) | runs org acted on — Global Threat Intelligence Lead for Influence Operations at Meta, signing its adversarial threat report of 7 April 2022. | first-hand 2022-04-07 |
| BEN NIMMO | leverage | 2 | public voice — Principal Investigator, Intelligence and Investigations at OpenAI; fronted OpenAI's June 2026 threat report on PRC-linked influence operations. | first-hand 2026-06-25 |
| BENJAMIN WEINSTEIN-RAUN | durability | 1 | publications — Co-author of Incomplete Tasks Induce Shutdown Resistance in Some Frontier LLMs (arXiv 2509.14260). | scholarly 2025-09-13 |
| BENJAMIN WEINSTEIN-RAUN | insulation | 2 | at will employee — Benjamin Weinstein-Raun employed-by AI Impacts | official |
| BENJAMIN WEINSTEIN-RAUN | insulation | 2 | at will employee — Benjamin Weinstein-Raun employed-by Palisade Research | official |
| BENJAMIN WEINSTEIN-RAUN | insulation | 2 | at will employee — Benjamin Weinstein-Raun employed-by Redwood Research | official |
| BENJAMIN WEINSTEIN-RAUN | insulation | 2 | at will employee — Listed as Sr Researcher at Palisade Research in the FY2024 Form 990 summary; no stated term. | official 2024-12-31 |
| BENJAMIN WEINSTEIN-RAUN | leverage | 6 | runs org acted on — Head of Research at Palisade Research and acting director of AI Impacts per Palisade's page. | first-hand |
| BENJAMIN WEINSTEIN-RAUN | leverage | 4 | advisory seat — Benjamin Weinstein-Raun advisory Survival and Flourishing Fund | official |
| BETH BARNES | durability | 6 | built operating — Founded METR (spun out of ARC Evals, announced 19 September 2023), still operating. | first-hand 2023-09-19 |
| BETH BARNES | durability | 4 | cited work — METR's HCAST and time-horizon evaluations, authored by her per METR's team page. | first-hand 2025-03-17 |
| BETH BARNES | insulation | 4 | appointed officer — Founder, CEO of METR, a 501(c)(3) with a board (page names Adam Gleave and Rajiv Dattani as advisor and board member); no source states her control. | first-hand 2023-12-04 |
| BETH BARNES | leverage | 6 | runs org acted on — Founder and CEO of METR, which runs pre-deployment evaluations of frontier models for labs and governments. | first-hand 2023-12-04 |
| BETH BARNES | leverage | 4 | advisory seat — Beth Barnes advisory LawZero | official |
| BIS / CBDC COMPLEX | entrenchment | 8 | program of record — The BIS balance sheet total was SDR 482 billion at 31 March 2026, built on deposits from the central banks that own it, with Swiss Headquarters Agreement immunity from execution on its assets. | first-hand 2026-03-31 |
| BIS / CBDC COMPLEX | entrenchment | 6 | operating — The BIS was chartered in 1930, is owned by 63 member central banks that account for about 95% of world GDP, and published its Annual Report 2025/26. | first-hand |
| BIS / CBDC COMPLEX | funding insulation | 4 | graded funding disclosed — The BIS states that currency deposits, primarily from central banks, constitute the largest share of its liabilities, with deposits of SDR 428 billion on a balance sheet of SDR 482 billion at 31 March 2026. | first-hand 2026-03-31 |
| BIS / CBDC COMPLEX | reach | 6 | output acted on — The BIS states that in 2025 the central banks of India, Indonesia, Malaysia, the Philippines, Singapore and Thailand incorporated Nexus Global Payments to bring Nexus to live implementation, and that the BIS has handed the project over to it. | first-hand 2025-01-01 |
| BIS / CBDC COMPLEX | reach | 4 | convenes or funds — The BIS Innovation Hub's Project Agora convenes eight central banks, including those of five major reserve currencies, for real-value testing of tokenised cross-border settlement. | first-hand 2026-05-27 |
| BLACKROCK | entrenchment | 8 | program of record — BlackRock states that through the Aladdin platform it provides technology solutions to over 1,000 clients, and its 10-K describes Aladdin assignments as typically long-term contracts providing recurring revenue. | first-hand 2024-06-30 |
| BLACKROCK | entrenchment | 6 | operating — BlackRock's FY2025 10-K reports total AUM of $14.0 trillion at 31 December 2025 and total 2025 revenue of $24,216 million. | first-hand 2025-12-31 |
| BLACKROCK | reach | 8 | contract or gate — BlackRock's FY2025 10-K states that Aladdin assignments are typically long-term contracts that provide recurring revenue, and that Aladdin represented the majority of its $2.0 billion technology services and subscription revenue. | first-hand |
| BLAKE BORGESON | durability | 6 | built operating — Blake Borgeson founded Recursion Pharmaceuticals | official |
| BLAKE BORGESON | insulation | 4 | appointed officer — Class III director of Recursion Pharmaceuticals with a term ending at the 2027 annual meeting; board since November 2013; proxy states no voting control. | official 2026-04-30 |
| BLAKE BORGESON | leverage | 4 | advisory seat — Blake Borgeson board-seat Machine Intelligence Research Institute (MIRI) | official |
| BLAKE BORGESON | leverage | 4 | advisory seat — Blake Borgeson board-seat Palisade Research | official |
| BLAKE BORGESON | leverage | 4 | advisory seat — Blake Borgeson board-seat Good Forever Foundation | official |
| BLAKE BORGESON | leverage | 4 | funds others — Donated $300,000 to MIRI (announced 5 August 2016). | first-hand 2016-08-05 |
| BRAD SMITH | durability | 6 | built operating — Co-founded Kids in Need of Defense, of which he is board chair emeritus per Microsoft's biography. | first-hand 2024-01-01 |
| BRAD SMITH | durability | 4 | cited work — Tools and Weapons (Penguin Press, 2019) and the May 2023 Microsoft AI governance blueprint under his byline. | first-hand 2019-09-01 |
| BRAD SMITH | insulation | 4 | appointed officer — Listed as Vice Chair and President among Microsoft's named executive officers in the October 2025 proxy. | official 2025-10-01 |
| BRAD SMITH | leverage | 8 | signs enforcement — Under his byline Microsoft announced it ceased and disabled services to a unit within the Israel Ministry of Defense after a terms-of-service review. | first-hand 2025-09-25 |
| BRAD SMITH | leverage | 4 | advisory seat — Brad Smith board-seat Netflix | official |
| BRAD SMITH | leverage | 4 | advisory seat — Brad Smith board-seat Princeton | official |
| BRI TREECE | durability | 6 | built operating — Bri Treece co-founder Fathom AI Inc | official |
| BRI TREECE | insulation | 4 | appointed officer — Co-Founder, President of Fathom; no source states control. | first-hand 2024-07-23 |
| BRI TREECE | insulation | 2 | at will employee — Bri Treece employed-by Center for AI Safety | official |
| BRI TREECE | leverage | 6 | runs org acted on — Co-Founder and President of Fathom, sponsor of California SB 813 on independent verification organizations. | first-hand 2024-07-23 |
| BRI TREECE | leverage | 4 | advisory seat — Bri Treece advisory AI Verification and Evaluation Research Institute (AVERI) | official |
| BRI TREECE | leverage | 4 | advisory seat — Bri Treece board-seat FAR.AI | official |
| BUCK SHLEGERIS | durability | 6 | built operating — Founder of Redwood Research, still operating. | official |
| BUCK SHLEGERIS | durability | 3 | cited work — Co-author of AI Control: Improving Safety Despite Intentional Subversion (arXiv 2312.06942). | scholarly 2023-12-12 |
| BUCK SHLEGERIS | durability | 1 | publications — Buck Shlegeris co-author Anthropic | scholarly |
| BUCK SHLEGERIS | insulation | 4 | appointed officer — CEO and one of three board members of Redwood Research; no source states majority control. | first-hand |
| BUCK SHLEGERIS | leverage | 6 | runs org acted on — CEO of Redwood Research, whose AI-control work Google DeepMind's roadmap (13 July 2026) thanks him for feedback on. | first-hand 2026-07-13 |
| BUCK SHLEGERIS | leverage | 4 | advisory seat — Buck Shlegeris advisory UK AI Security Institute | official |
| BUCK SHLEGERIS | leverage | 4 | advisory seat — Buck Shlegeris board-seat Apart Research | official |
| BUCK SHLEGERIS | leverage | 4 | advisory seat — Buck Shlegeris board-seat Alignment Research Center | official |
| BUCK SHLEGERIS | leverage | 4 | convenes — Announced ControlConf 2026, a two-day AI control conference in Berkeley, 18 to 19 April 2026. | first-hand 2026-02-25 |
| CAMILLE FRANÇOIS | durability | 6 | built operating — Founded ROOST (launched February 2025, $27M+), co-released gpt-oss-safeguard with OpenAI on 29 October 2025. | first-hand 2025-10-29 |
| CAMILLE FRANÇOIS | durability | 1 | publications — Camille François co-author Ben Nimmo | scholarly |
| CAMILLE FRANÇOIS | insulation | 4 | appointed officer — President of ROOST and member of its board, which stewards long-term strategy (inaugural board 22 December 2025); no source states control. | first-hand 2025-12-22 |
| CAMILLE FRANÇOIS | insulation | 1 | at will employee — Camille François employed-by Columbia | scholarly |
| CAMILLE FRANÇOIS | insulation | 1 | at will employee — Camille François employed-by Niantic | scholarly |
| CAMILLE FRANÇOIS | insulation | 1 | at will employee — Camille François employed-by Google / Alphabet | press |
| CAMILLE FRANÇOIS | insulation | 1 | at will employee — Camille François employed-by Graphika | press |
| CAMILLE FRANÇOIS | leverage | 6 | runs org acted on — Founder and president of ROOST, whose Osprey rules engine is used by Bluesky, Discord and Matrix and whose tools ROOST says protect over 370 million monthly users. | first-hand 2026-10-05 |
| CAMILLE FRANÇOIS | leverage | 5 (ended) | runs org acted on — Chief Innovation Officer of Graphika (named January 2019), which led the Election Integrity Partnership's Graphika team. | press 2019-01-27 |
| CARI TUNA | durability | 6 | built operating — Coefficient Giving's governance page names Cari Tuna Chair of the Board of Managers of Coefficient Giving LLC (formerly Open Philanthropy) and of the Board of Directors of Coefficient Giving Research. | first-hand |
| CARI TUNA | durability | 6 | built operating — Cari Tuna founded Good Ventures | official |
| CARI TUNA | durability | 6 | built operating — Good Ventures governance page lists Cari Tuna as Chair of the foundation she co-founded in 2011; the foundation is a private foundation that makes grants and still operates. | first-hand |
| CARI TUNA | insulation | 6 | independent money — Good Ventures, which she co-founded in 2011 and chairs, is the couple's own giving vehicle; Coefficient states she and Moskovitz have given away more than $4 billion, so her means do not depend on the evaluators and labs the money funds. | first-hand |
| CARI TUNA | leverage | 4 | advisory seat — Cari Tuna board-seat Open Philanthropy | official |
| CARI TUNA | leverage | 4 | advisory seat — Cari Tuna board-seat GiveWell | official |
| CARI TUNA | leverage | 4 | funds others — Coefficient Giving's page on Cari Tuna states she is co-founder and Chair of Good Ventures, which she and Dustin Moskovitz created in 2011, and that they have given away more than $4 billion. | first-hand |
| CENTER FOR COUNTERING DIGITAL HATE | entrenchment | 6 | operating — CCDH's calendar 2025 Form 990 reports nine employees and net assets of $1,866,587, and on 15 October 2026 the organization announced it was renamed the Center for Countering Digital Harm. | first-hand 2026-10-15 |
| CENTER FOR COUNTERING DIGITAL HATE | funding insulation | 6 | independent funding — CCDH's calendar 2025 Form 990 reports total revenue of $3,526,306, all but $32,244 of it contributions and grants, no program-service revenue, and USAspending lists no federal award to it. | official |
| CENTER FOR COUNTERING DIGITAL HATE | reach | 6 | output acted on — Meta stated it removed over three dozen Pages, groups and accounts linked to the 12 people named in CCDH's Disinformation Dozen report, including at least one linked to each of the 12, while disputing the report's 65 percent figure. | official 2021-08-18 |
| CELLEBRITE | entrenchment | 8 | program of record — USAspending shows Cellebrite Inc. receiving annual ICE purchase orders in the same series since 2019-20, from $2.9 million that year to $12,794,860 for 2026-27. | official 2026-09-15 |
| CELLEBRITE | entrenchment | 6 | operating — Cellebrite's 2025 Form 20-F reports total revenue of $475.7 million and approximately 7,000 customers, about 5,300 of them federal, state and local agencies. | first-hand 2026-03-03 |
| CELLEBRITE | funding insulation | 4 | graded funding disclosed — Cellebrite's 2025 Form 20-F states that Public Sector Customers accounted for more than 90% of total revenue in 2025 and US federal customers for approximately 16%. | first-hand 2026-03-03 |
| CELLEBRITE | reach | 8 | contract or gate — USAspending records a one-year ICE delivery order, 70CMSD26FR0000130, signed 15 September 2026 for $12,794,860 for Cellebrite products and services supporting homeland security investigations with digital forensic capabilities. | official 2026-09-15 |
| CENTER FOR AI SAFETY | entrenchment | 6 | operating — ProPublica lists Center For Artificial Intelligence Safety Inc, San Francisco, EIN 88-1751310, with tax exemption issued January 2023 and Form 990 filings for FY2022 through FY2024. | official 2023-01-01 |
| CENTER FOR AI SAFETY | funding insulation | 6 | independent funding — ProPublica's summary of CAIS's Form 990 data gives FY2024 total revenue of $10,235,085, of which $9,596,669 is contributions; no payment from any AI lab is documented. | official |
| CENTER FOR AI SAFETY | reach | 4 | convenes or funds — CAIS organized and hosts the 30 May 2023 Statement on AI Risk, signed by Geoffrey Hinton, Yoshua Bengio, Demis Hassabis, Sam Altman and Dario Amodei, and its Action Fund was a named sponsor of California's SB 1047. | first-hand 2023-05-30 |
| CENTER FOR HUMANE TECHNOLOGY | entrenchment | 6 | operating — CHT's Board announced on 1 October 2026 that it will wind down its in-house policy, litigation, research and other programs and keep founder-led public communication and decision-maker influence, with the Stop the AI Drift campaign continuing through the end of October. | first-hand 2026-10-01 |
| CENTER FOR HUMANE TECHNOLOGY | funding insulation | 6 | independent funding — CHT's FY2025 Form 990 reports revenue of $5,785,317 with contributions and grants of $5,401,973 (93.4 percent), no government grants, and a Schedule B that names no donor. | official |
| CENTER FOR HUMANE TECHNOLOGY | reach | 4 (ended) | convenes or funds — CHT's FY2024 Schedule I lists grants out of $395,000: Hopewell Fund $220,000, Campaign for Accountability $75,000, Moms Against Media Addiction $75,000 and Ben Franklin Society Library $25,000. | official |
| CHARLEMAGNE LABS | entrenchment | 6 | operating — Charlemagne Labs sells Agent Charley, soft-launched 31 October 2025, lists open roles on its careers page, and announced on 5 October 2026 that it is a finalist in the Ninth Annual DataTribe Challenge. | first-hand 2026-10-05 |
| CHARLEMAGNE LABS | reach | 6 | output acted on — Meta's Muse Spark 1.1 Evaluation Report assesses Cyber 3 through a multi-turn social-engineering benchmark developed by Charlemagne Labs and prints composite scores for Muse Spark 1.1, GPT-5.5, Claude Opus 4.8 and Gemini 3.1 Pro. | official 2026-07-09 |
| CHARLEMAGNE LABS | reach | 6 | output acted on — Charlemagne Labs evaluator-for Meta | official |
| CHARLES KOCH FOUNDATION | entrenchment | 8 | program of record — The Charles Koch Foundation publishes 54 major multi-year university grant agreements, among them the 2019 New York University agreement committing up to $5,000,000 in annual instalments over six requests from 2019 to 2024. | first-hand 2019-09-01 |
| CHARLES KOCH FOUNDATION | entrenchment | 6 | operating — The Charles Koch Foundation has filed Form 990-PF for every year from 2002 to 2024 and reported $67,819,528 in charitable disbursements for 2024. | official |
| CHARLES KOCH FOUNDATION | funding insulation | 8 | endowment or appropriation — The Charles Koch Foundation's 2024 Form 990-PF reports total assets of $758,930,141 and contributions received of $44,958, against charitable disbursements of $67,819,528. | official |
| CHARLES KOCH FOUNDATION | reach | 4 | convenes or funds — The Charles Koch Foundation's 2024 Form 990-PF shows $64,073,452 paid in grants during the year, the largest $23,675,637 to the George Mason University Foundation. | first-hand |
| CHARLES KOCH FOUNDATION | reach | 4 | convenes or funds — NYU Center for Social Media and Politics funded-by Charles Koch Institute / Charles Koch Foundation | official |
| CHARLES KOCH FOUNDATION | reach | 4 | convenes or funds — Kate Klonick funded-by Charles Koch Institute / Charles Koch Foundation | official |
| CHARLES TEAGUE | durability | 6 | built operating — Charles Teague founded Meridian Labs | official |
| CHARLES TEAGUE | durability | 6 | built operating — Meridian Labs, co-founded by Teague in 2025, maintains Inspect AI and since 7 May 2026 Petri, which Anthropic handed over to it. | first-hand 2026-05-07 |
| CHARLES TEAGUE | durability | 4 | cited work — The AEF-1 standard (Version 1, updated December 4, 2025) lists Charles Teague among its authors. | first-hand 2025-12-04 |
| CHARLES TEAGUE | durability | 2 | publications — Charles Teague co-author Beth Barnes | official |
| CHARLES TEAGUE | durability | 2 | publications — Charles Teague co-author Chris Painter | official |
| CHARLES TEAGUE | durability | 2 | publications — Charles Teague co-author Jaime Sevilla | official |
| CHARLES TEAGUE | durability | 2 | publications — Charles Teague co-author Miles Brundage | official |
| CHARLES TEAGUE | durability | 2 | publications — Charles Teague co-author J.J. Allaire | official |
| CHARLES TEAGUE | durability | 2 | publications — Charles Teague co-author Kai Fronsdal | official |
| CHARLES TEAGUE | insulation | 4 | appointed officer — Meridian Labs' team page lists Charles Teague as CEO and co-founder of the nonprofit. | first-hand |
| CHARLES TEAGUE | insulation | 2 | at will employee — Charles Teague employed-by Posit PBC | official |
| CHARLES TEAGUE | insulation | 2 | at will employee — Charles Teague employed-by RAND | official |
| CHARLES TEAGUE | leverage | 6 | runs org acted on — Meridian Labs' team page lists Charles Teague as CEO and co-founder; its About page says Inspect AI, built by Meridian's founding team with the UK AISI, is the standard framework for UK AISI, US CAISI, EU AI Office, METR, Apollo, RAND and others. | first-hand |
| CHARLOTTE WILLNER | durability | 6 | built operating — The Internet Education Foundation announcement of 19 Nov 2020 names her founding Executive Director of TSPA (501(c)(6)) and the Trust & Safety Foundation; TSPA's own pages show both still operating in 2026. | first-hand 2020-11-19 |
| CHARLOTTE WILLNER | insulation | 4 | appointed officer — TSPA's team page lists Charlotte Willner as Executive Director; its board page lists Del Harvey as Board Chair. | first-hand |
| CHARLOTTE WILLNER | leverage | 4 | convenes — TSPA's September 2022 release, posted under Executive Director Charlotte Willner, announced TrustCon, the first global trust and safety conference; TSPA's 2026 recap counts more than 1,350 attendees from nearly 450 companies at the fifth. | first-hand 2026-08-18 |
| CHINA SURVEILLANCE BLOC (HIKVISION · DAHUA · SENSETIME) | entrenchment | 6 | operating — Hikvision's 2024 annual report is the filing of an operating listed company (002415.SZ) with CETC as actual controller and 36.93% shareholder through China Electronics Technology HIK Group. | first-hand 2025-04-26 |
| CHINA SURVEILLANCE BLOC (HIKVISION · DAHUA · SENSETIME) | funding insulation | 6 | independent funding — Hikvision's 2024 annual report shows sales to its top five customers at RMB 2,343,413,902, 2.53% of total sales, so no customer approaches 20% of revenue. | first-hand 2025-04-26 |
| CHRIS LEHANE | durability | 2 | publications — OpenAI publishes the letters he signs and a 9 September 2026 policy post bylined to him as Chief Global Affairs Officer. | first-hand 2026-09-09 |
| CHRIS LEHANE | insulation | 4 | appointed officer — OpenAI's letters name Chris Lehane as Chief Global Affairs Officer, a senior executive of the company. | first-hand 2025-08-11 |
| CHRIS LEHANE | leverage | 2 | public voice — OpenAI's 11 August 2025 letter to Governor Newsom is from Chris Lehane, Chief Global Affairs Officer, recommending California treat frontier developers as compliant when they sign the EU Code of Practice or agreements with CAISI. | first-hand 2025-08-11 |
| CHRIS MESEROLE | durability | 2 | publications — Brookings maintains his expert page; he previously directed its Artificial Intelligence and Emerging Technology Initiative. | first-hand |
| CHRIS MESEROLE | insulation | 4 | appointed officer — Anthropic, Google, Microsoft and OpenAI announced the selection of Chris Meserole as the Forum's first Executive Director, with the creation of a $10 million+ AI Safety Fund. | first-hand 2023-10-25 |
| CHRIS MESEROLE | leverage | 4 | convenes — The Frontier Model Forum states it is led by Executive Director Chris Meserole and overseen by an operating board of representatives of its member organizations (Anthropic, Google, Microsoft, OpenAI and others). | first-hand |
| CHRIS MESEROLE | leverage | 2 | public voice — Chris Meserole testified U.S. House of Representatives | official |
| CHRIS OLAH | durability | 6 | built operating — Chris Olah co-founder Anthropic | official |
| CHRIS OLAH | durability | 4 | cited work — Anthropic's constitution page credits Chris Olah, Jared Kaplan and Holden Karnofsky with 'significant contributions to its content and development'. | first-hand |
| CHRIS OLAH | durability | 2 | publications — Chris Olah co-author Joe Carlsmith | official |
| CHRIS OLAH | durability | 2 | publications — Chris Olah co-author Holden Karnofsky | official |
| CHRIS OLAH | durability | 1 | publications — Chris Olah co-author Amanda Askell | scholarly |
| CHRIS OLAH | durability | 1 | publications — Chris Olah co-author Dario Amodei | scholarly |
| CHRIS OLAH | durability | 1 | publications — Chris Olah co-author Jared Kaplan | scholarly |
| CHRIS OLAH | insulation | 2 | at will employee — Chris Olah employed-by Google / Alphabet | official |
| CHRIS OLAH | insulation | 2 | at will employee — Chris Olah employed-by OpenAI | official |
| CHRIS OLAH | leverage | 6 | runs org acted on — Olah says 'I lead a research team that studies the internal structure of these models'; Anthropic's August 2026 Risk Report says interpretability evidence is used in pre-deployment review of its models. | first-hand 2026-05-25 |
| CHRIS PAINTER | durability | 4 | cited work — The AEF-1 standard (Version 1, December 4, 2025) lists Christopher Painter among its authors. | first-hand 2025-12-04 |
| CHRIS PAINTER | durability | 4 | cited work — METR's Common Elements page says the Frontier Safety Policy concept was introduced by METR in 2023 and lists twelve developers' policies; his METR bio says he helped shape them. | first-hand |
| CHRIS PAINTER | insulation | 4 | appointed officer — METR lists him as President beside Beth Barnes (Founder, CEO); no control or term is stated. | first-hand |
| CHRIS PAINTER | insulation | 2 | at will employee — Chris Painter employed-by DoD | official |
| CHRIS PAINTER | insulation | 2 | at will employee — Chris Painter employed-by METR (Model Evaluation & Threat Research) | official |
| CHRIS PAINTER | leverage | 6 | runs org acted on — METR's team page lists Chris Painter as President, who 'leads engagement with governments and AI labs' and 'has helped shape industry Frontier Safety Policies'; METR says the concept it introduced in 2023 now has twelve published company policies. | first-hand |
| CHRIS PAINTER | leverage | 2 | public voice — METR President Chris Painter testified before the Senate HSGAC subcommittee on 30 September 2026 in 'Rogue AI: Securing the Homeland Against AI Agent Attacks'. | first-hand 2026-09-30 |
| CHRIS PAINTER | leverage | 2 | public voice — Chris Painter testified US Senate | official |
| CISA | entrenchment | 10 | statute or infrastructure — Pub. L. 115-278 section 2202 establishes the Cybersecurity and Infrastructure Security Agency within DHS, enacted 16 November 2018. | official 2018-11-16 |
| CISA | funding insulation | 8 | endowment or appropriation — DHS's FY2027 budget justification shows CISA at $2,873,008 thousand in the FY2025/FY2026 continuing-resolution baseline and $2,487,194 thousand in the FY2027 request. | official 2026-04-03 |
| CISA | reach | 10 | statutory power — CISA Binding Operational Directive 26-04, issued 10 June 2026, is described by CISA as a compulsory direction to federal civilian executive branch agencies. | official 2026-06-10 |
| CLARA TSAO | durability | 6 | built operating — TSPA's board page names her Co-Founder of the Trust & Safety Professional Association, which still operates. | first-hand |
| CLARA TSAO | insulation | 4 | appointed officer — TSPA's board page lists her as Co-Founder and Board Member of the association. | first-hand |
| CLARA TSAO | insulation | 2 | at will employee — Clara Tsao employed-by US Department of Homeland Security | official |
| CLARA TSAO | leverage | 4 | advisory seat — TSPA's board page lists Clara Tsao as Co-Founder and Board Member and as a senior advisor to the UN-backed Tech Against Terrorism. | first-hand |
| CLEARVIEW AI | entrenchment | 8 | program of record — ICE award 70CMSD25P00000111 is a multi-year federal contract running to 4 September 2027 with a $9,225,000 ceiling. | official 2025-09-05 |
| CLEARVIEW AI | entrenchment | 6 | operating — The Seventh Circuit (No. 25-1673, 13 July 2026) describes Clearview AI, Inc. as operating a search engine for faces built from scraped public photographs. | official 2026-07-13 |
| CLEARVIEW AI | reach | 8 | contract or gate — ICE awarded Clearview AI, Inc. contract 70CMSD25P00000111 for facial recognition software for Homeland Security Investigations, $7,687,500 obligated, performance 5 September 2025 to 4 September 2027. | official 2025-09-05 |
| CLEARVIEW AI | reach | 6 | output acted on — Clearview AI contract US Immigration and Customs Enforcement | official |
| CLEARVIEW AI | reach | 6 | output acted on — Clearview AI contract US Customs and Border Protection | official |
| CLEARVIEW AI | reach | 6 | output acted on — Clearview AI contract Federal Bureau of Investigation | official |
| CLÉMENT DELANGUE | durability | 6 | built operating — Hugging Face, the company he co-founded and leads as CEO, filed a Form D in 2019 and is the subject of Nvidia's 2 September 2026 definitive acquisition agreement, so it still operates. | official 2026-09-02 |
| CLÉMENT DELANGUE | insulation | 4 | appointed officer — Hugging Face's SEC Form D (2019) lists Clément Delangue as an executive officer and director; Nvidia's 8-K records a definitive agreement to acquire Hugging Face and states no voting control by him. | official 2019-12-12 |
| CLÉMENT DELANGUE | leverage | 4 | funds others — Artificial Analysis funded-by Clément Delangue | official |
| CLÉMENT DELANGUE | leverage | 2 | public voice — Hugging Face's July 2026 incident disclosure states the intrusion was driven end to end by an autonomous AI agent system and that hosted providers' guardrails blocked its forensic requests; he briefed the UN Security Council on 23 September 2026 as Co-Founder and CEO. | first-hand 2026-07-16 |
| CLÉMENT DELANGUE | leverage | 2 | public voice — Clément Delangue testified U.S. House of Representatives | official |
| COLLECTIVE INTELLIGENCE PROJECT | entrenchment | 6 | operating — ProPublica lists Collective Intelligence Project Inc, San Francisco, EIN 92-0327339, as a tax-exempt organization (exemption issued October 2024) with filed returns. | official 2024-10-01 |
| COLLECTIVE INTELLIGENCE PROJECT | funding insulation | 6 | independent funding — CIP states it operates exclusively through philanthropic charitable donations and does not accept funding from commercial partnerships or contractual relationships, naming funders including RWJF, Google.org, Ford Foundation, Omidyar Network and the Survival and Flourishing Fund. | first-hand |
| COLLECTIVE INTELLIGENCE PROJECT | reach | 6 | output acted on — Anthropic states it partnered with the Collective Intelligence Project to run a public input process and trained a model against the resulting public constitution. | first-hand 2023-10-17 |
| CONRAD STOSZ | durability | 4 | cited work — The AEF-1 standard (Version 1, December 4, 2025) names Conrad Stosz as its correspondence author. | first-hand 2025-12-04 |
| CONRAD STOSZ | insulation | 4 | appointed officer — Transluce announced on 22 October 2025 it named Conrad Stosz its first Head of Governance, to lead efforts to define standards and best practices for AI evaluations. | first-hand 2025-10-22 |
| CONRAD STOSZ | insulation | 2 | at will employee — Conrad Stosz employed-by DoD | official |
| CONRAD STOSZ | insulation | 2 | at will employee — Conrad Stosz employed-by OMB | official |
| CONRAD STOSZ | insulation | 2 | at will employee — Conrad Stosz employed-by Center for AI Standards and Innovation (NIST) | official |
| CONRAD STOSZ | insulation | 2 | at will employee — Conrad Stosz employed-by Transluce | official |
| CONRAD STOSZ | leverage | 4 | advisory seat — Conrad Stosz board-seat AI Evaluator Forum | official |
| CONRAD STOSZ | leverage | 4 | convenes — The AI Evaluator Forum's members page lists Conrad Stosz as Chair of a network of eight independent evaluators including METR, RAND, Meridian Labs and Transluce. | first-hand |
| DAN GUIDO | durability | 6 | built operating — Trail of Bits' About page dates the firm's founding to 2012 and records its ongoing work, including DARPA's AI Cyber Challenge. | first-hand 2012-01-01 |
| DAN GUIDO | insulation | 4 | appointed officer — Trail of Bits' team page lists Dan Guido, CEO and co-founder, alongside Alex Sotirov, Co-CEO; no control is stated. | first-hand |
| DAN GUIDO | leverage | 6 | runs org acted on — Trail of Bits' team page lists Dan Guido as CEO and co-founder; its library lists OpenAI and Google as clients of AI security reviews (e.g. the OpenAI OpenClaw review). | first-hand |
| DAN GUIDO | leverage | 4 (ended) | advisory seat — The CFTC release of 14 July 2023 names Dan Guido co-chair of the Technology Advisory Committee's Subcommittee on Cybersecurity; the committee's current roster page no longer lists him. | official 2023-07-14 |
| DAN HENDRYCKS | durability | 6 | built operating — The Center for AI Safety's About page lists Hendrycks as 'Executive & Research Director' and says the center has 'provided access to our computing infrastructure to >100 researchers' since March 2023. | first-hand |
| DAN HENDRYCKS | durability | 3 | cited work — Humanity's Last Exam (arXiv:2501.14249, submitted 24 January 2025, revised to v12 on 19 October 2026) lists Dan Hendrycks as last author. | scholarly 2025-01-24 |
| DAN HENDRYCKS | insulation | 4 | appointed officer — Form 990 data on ProPublica list Daniel James Hendrycks as 'Executive Dir.' of Center For Artificial Intelligence Safety Inc at $314,534 for FY2024 (President in FY2022; exemption issued January 2023). | official |
| DAN HENDRYCKS | leverage | 6 | runs org acted on — xAI's Grok 4 release page reports Grok 4 Heavy scoring 50% on Humanity's Last Exam, the benchmark whose arXiv paper lists Hendrycks as last author. | official |
| DAN HENDRYCKS | leverage | 4 | advisory seat — Dan Hendrycks advisory Gray Swan AI | official |
| DAN LAHAV | durability | 5 | built operating — Irregular, co-founded by Lahav (TechCrunch), still operates and supplies cyber evaluations to frontier labs. | press 2025-09-17 |
| DAN LAHAV | durability | 3 | cited work — RAND's 'Securing AI Model Weights' (30 May 2024) lists Dan Lahav as an author; it identifies 38 attack vectors and defines five security levels. | scholarly 2024-05-30 |
| DAN LAHAV | insulation | 3 | appointed officer — TechCrunch (17 Sept 2025) identifies Dan Lahav as co-founder and CEO of Irregular, which raised $80 million led by Sequoia and Redpoint. | press 2025-09-17 |
| DAN LAHAV | leverage | 5 | runs org acted on — Google DeepMind's Gemini 4 safety report says its cyber CCL assessment rests on CyScenarioBench, co-developed with Irregular, the company Dan Lahav co-founded and leads as CEO (TechCrunch); OpenAI's GPT-6 Astra page carries an Irregular external cyber evaluation section. | press |
| DANIEL KOKOTAJLO | durability | 6 | built operating — The AI Futures Project, a 501(c)(3) he leads and which published AI 2027, operates and lists him as Executive Director. | first-hand |
| DANIEL KOKOTAJLO | insulation | 4 | appointed officer — The AI Futures Project's About page lists Daniel Kokotajlo as Executive Director of a 501(c)(3) nonprofit; no term or removal protection is stated. | first-hand |
| DANIEL KOKOTAJLO | insulation | 2 | at will employee — Daniel Kokotajlo employed-by Apollo Research | official |
| DANIEL KOKOTAJLO | insulation | 2 | at will employee — Daniel Kokotajlo employed-by OpenAI | official |
| DANIEL KOKOTAJLO | leverage | 4 | advisory seat — Daniel Kokotajlo advisory Epoch AI | official |
| DANIEL KOKOTAJLO | leverage | 2 | public voice — Daniel Kokotajlo testified US Senate | official |
| DARIO AMODEI | durability | 6 | built operating — Amodei writes 'When we founded Anthropic, we decided to make [interpretability] a central part of the new company's direction' and calls Chris Olah his 'co-founder'. | first-hand 2025-04-01 |
| DARIO AMODEI | insulation | 4 | appointed officer — Anthropic's company page lists Dario Amodei on the Board of Directors and says 'Our Board of Directors is elected by stockholders and our Long-Term Benefit Trust'. | first-hand |
| DARIO AMODEI | leverage | 8 | signs deployment — Anthropic's Responsible Scaling Policy v2.2 says Safeguards Reports are escalated to the CEO and the Responsible Scaling Officer, who 'make the ultimate determination as to whether we have satisfied the Required Safeguards and decide any deployment-related issues' (same for the Capability Report). | first-hand 2025-05-14 |
| DAVE WILLNER | durability | 6 | built operating — Dave Willner co-founder Zentropi | official |
| DAVE WILLNER | leverage | 7 (ended) | signs enforcement — He was OpenAI's head of trust and safety for about a year and a half, stepping down in July 2023 to an advisory role. | press 2023-07-21 |
| DAVE WILLNER | leverage | 6 | runs org acted on — Berkman Klein Center page calls him co-founder of Zentropi, whose site says customers create and deploy custom content labelers and lists Character.AI, Runway, Stanford, Koko and Cantina as trusted by. | official 2026-10-05 |
| DAVID SACKS | durability | 6 | built operating — David Sacks co-founder Craft Ventures | official |
| DAVID SACKS | durability | 4 (ended) | cited work — America's AI Action Plan, which the January 2025 order tasked the Special Advisor for AI and Crypto with developing, carries his signature block. | official 2025-07-23 |
| DAVID SACKS | durability | 2 | publications — David Sacks co-author Michael Kratsios | official |
| DAVID SACKS | insulation | 4 | appointed officer — Appointed co-chair of PCAST by the President, an advisory body whose members serve at the appointing authority's pleasure. | official 2026-03-25 |
| DAVID SACKS | leverage | 8 (ended) | holds policy pen — EO 14365 directs the Special Advisor for AI and Crypto with the APST to jointly prepare a legislative recommendation for a uniform federal AI policy framework; the January 2025 AI order tasks the seat with developing the AI Action Plan. He left the seat in March 2026. | official 2025-12-11 |
| DAVID SACKS | leverage | 4 | advisory seat — The White House announced Sacks and Michael Kratsios as co-chairs of the President's Council of Advisors on Science and Technology. | official 2026-03-25 |
| DEAN BALL | durability | 4 (ended) | cited work — FAI credits his co-authorship of the White House AI Action Plan; he left the OSTP seat in August 2025. | first-hand 2026-06-18 |
| DEAN BALL | durability | 2 | publications — Dean Ball co-author AI Verification and Evaluation Research Institute (AVERI) | official |
| DEAN BALL | durability | 2 | publications — Hyperdimensional archive of essays including On Private Governance and What Should Be Done remains published. | first-hand 2026-06-26 |
| DEAN BALL | insulation | 2 | at will employee — Dean Ball employed-by OpenAI | official |
| DEAN BALL | leverage | 4 | advisory seat — Dean Ball advisory Fathom AI Inc | official |
| DEAN BALL | leverage | 2 | public voice — Author of the Hyperdimensional newsletter on AI policy since January 2024, and announced he joins OpenAI as Head of Strategic Futures, a team shaping frontier AI policy inside the company. | first-hand 2026-06-18 |
| DEL HARVEY | durability | 6 (ended) | built operating — TSPA's board page says she founded, built and led the global T&S team at Twitter for its first 13 years. | first-hand |
| DEL HARVEY | leverage | 8 (ended) | signs enforcement — Led Trust and Safety at Twitter for its first 13 years and was Vice President of Trust and Safety by 2016; she left in late 2021. | first-hand 2021-12-31 |
| DEL HARVEY | leverage | 4 | advisory seat — Board Member and Board Chair of the Trust & Safety Professional Association. | first-hand 2025-06-25 |
| DELOITTE (AI ASSURANCE AND TESTING) | entrenchment | 8 | program of record — CMS awarded Deloitte Consulting LLP order 75FCMC26F0061 for a dedicated AI research practice, $5,675,345.88 obligated, ceiling $28,234,945.08, performance 17 April 2026 to 16 April 2027 with potential end 16 April 2031. | official 2026-04-17 |
| DELOITTE (AI ASSURANCE AND TESTING) | funding insulation | 6 | independent funding — Deloitte reports aggregate global revenue of US$74.5 billion for the fiscal year ending 31 May 2026, spread across a client base in which no single frontier-lab customer is stated to be near 20 percent. | first-hand 2026-09-01 |
| DELOITTE (AI ASSURANCE AND TESTING) | reach | 6 | output acted on — Deloitte Consulting contract US Department of Health and Human Services | official |
| DELOITTE (AI ASSURANCE AND TESTING) | reach | 6 | output acted on — Deloitte Consulting contract DoD | official |
| DELOITTE (AI ASSURANCE AND TESTING) | reach | 6 | output acted on — Deloitte Consulting evaluator-for Meta | official |
| DELOITTE (AI ASSURANCE AND TESTING) | reach | 6 | output acted on — Deloitte Consulting evaluator-for Anthropic | official |
| DELOITTE (AI ASSURANCE AND TESTING) | reach | 6 | output acted on — Anthropic's Claude Opus 4.5 system card states its long-form virology evaluations were developed with SecureBio, Deloitte and Signature Science. | first-hand 2025-11-24 |
| DELOITTE (AI ASSURANCE AND TESTING) | reach | 5 | output acted on — Deloitte Consulting evaluator-for Amazon | scholarly |
| DEMIS HASSABIS | durability | 6 | built operating — Google's author page states 'Demis Hassabis is the co-founder and Chair of Google DeepMind, one of the world's leading AI research groups.' | official |
| DEMIS HASSABIS | durability | 6 | built operating — Demis Hassabis founded Isomorphic Labs | official |
| DEMIS HASSABIS | durability | 4 | cited work — The Nobel Foundation awarded half of the 2024 Chemistry prize jointly to Demis Hassabis and John Jumper of Google DeepMind 'for protein structure prediction' (not re-fetched). | official 2024-10-09 |
| DEMIS HASSABIS | durability | 2 | publications — Demis Hassabis co-author James Manyika | official |
| DEMIS HASSABIS | insulation | 4 | appointed officer — Google's message states Hassabis 'will become the Chair of GDM and Chief Scientist of Alphabet' and that he handed over 'my day-to-day operational responsibilities at GDM' to Koray Kavukcuoglu, SVP of Google DeepMind. | official 2026-08-05 |
| DEMIS HASSABIS | insulation | 2 | at will employee — Demis Hassabis employed-by Alphabet Inc. | official |
| DEMIS HASSABIS | leverage | 6 | runs org acted on — Google's 5 August 2026 message says 'Demis will become the Chair of GDM and Chief Scientist of Alphabet' while continuing to lead Isomorphic Labs, with Koray Kavukcuoglu leading GDM day to day. | official 2026-08-05 |
| ATLANTIC COUNCIL DFRLAB | entrenchment | 6 | operating — DFRLab's own about page describes it as incubated at the Atlantic Council in 2016, and its home page feed carries items to 12 October 2026. | first-hand 2026-10-12 |
| ATLANTIC COUNCIL DFRLAB | funding insulation | 6 | independent funding — DFRLab's parent, the Atlantic Council, reports FY2024 total revenue of $69,901,633 of which government grants are $9,908,972, with 35 of 36 voting board members independent. | official 2025-12-01 |
| ATLANTIC COUNCIL DFRLAB | reach | 6 | output acted on — Facebook announced that Atlantic Council DFRLab experts would work with its security, policy and product teams to give it real-time insights on emerging threats and disinformation campaigns. | first-hand 2018-05-17 |
| DREADNODE | entrenchment | 6 | operating — Dreadnode's own research page shows it operating and publishing a benchmark data release on 8 October 2026. | first-hand 2026-10-08 |
| DREADNODE | funding insulation | 6 | independent funding — Dreadnode announced a $14 million Series A on 25 February 2025 led by Decibel with Next Frontier Capital, In-Q-Tel, Sands Capital and Indie VC; investors are venture funds, not the labs whose models it grades. | first-hand 2025-02-25 |
| DREADNODE | reach | 2 | publishes — Dreadnode publishes DreadIndex, a benchmark it describes as designed to evaluate the offensive cyber security capabilities of language models (data release 8.1, 8 October 2026). | first-hand 2026-10-08 |
| DUSTIN MOSKOVITZ | durability | 6 | built operating — Co-founded Good Ventures, which describes itself as having distributed more than $7 billion, and which operates. | first-hand |
| DUSTIN MOSKOVITZ | insulation | 10 | voting control — Asana's 2026 proxy statement is said to list him at 74.2 percent of total voting power through Class B shares carrying 10 votes each (table not confirmed on re-read). | official 2026-04-20 |
| DUSTIN MOSKOVITZ | leverage | 8 | signs grants — Sits on the board (Secretary and Treasurer) of Good Ventures, a private foundation that makes grants; its FY2025 990-PF reports $306,706,819 in grants and his trust as the main contributor. | first-hand |
| DUSTIN MOSKOVITZ | leverage | 4 | funds others — Open Philanthropy funded-by Dustin Moskovitz | official |
| DUSTIN MOSKOVITZ | leverage | 4 | funds others — Good Ventures funded-by Dustin Moskovitz | official |
| DUSTIN MOSKOVITZ | leverage | 4 | funds others — Survival and Flourishing Fund funded-by Dustin Moskovitz | official |
| DUSTIN MOSKOVITZ | leverage | 4 | funds others — Resolution funded-by Dustin Moskovitz | official |
| DUSTIN MOSKOVITZ | leverage | 4 | funds others — Anthropic funded-by Dustin Moskovitz | official |
| DUSTIN MOSKOVITZ | leverage | 3 | advisory seat — Dustin Moskovitz advisory Anthropic | press |
| EGE ERDIL | durability | 6 | built operating — Ege Erdil co-founder Tamay Besiroglu | official |
| EGE ERDIL | durability | 6 | built operating — Ege Erdil founded Mechanize | official |
| EGE ERDIL | durability | 6 | built operating — Mechanize, which he co-founded in April 2025, is a team of around 35 people building environments and evals. | first-hand 2025-04-17 |
| EGE ERDIL | durability | 3 (ended) | cited work — Second author of the FrontierMath benchmark paper. | scholarly 2024-11-07 |
| EGE ERDIL | insulation | 2 | at will employee — Ege Erdil employed-by Epoch AI | official |
| EGE ERDIL | leverage | 6 | runs org acted on — Co-founder of Mechanize, whose environments are used by frontier AI labs to train models or measure their capabilities; its announcement lists Barnett, Besiroglu and Erdil as authors/founders. | first-hand 2026-04-24 |
| eIDAS 2.0 / EU DIGITAL IDENTITY WALLET (EUDI) | entrenchment | 10 | statute or infrastructure — Regulation (EU) 2024/1183 amending eIDAS is in force, with the Commission's letter applying its Articles 5a, 5b and 5f as binding law on wallet issuers, relying parties and very large platforms. | official 2026-04-20 |
| eIDAS 2.0 / EU DIGITAL IDENTITY WALLET (EUDI) | reach | 10 | statutory power — The Commission's DG CNECT reply of 20 April 2026 states that Article 5f(3) of Regulation (EU) 2024/1183 obliges very large online platforms to accept and facilitate the wallet for user authentication on the user's request, and that Article 5b(3) binds relying parties in all Member States. | official 2026-04-20 |
| ELECTION INTEGRITY PARTNERSHIP | entrenchment | 2 (ended) | reports persist — The EIP's home page states it was founded in 2020 and finished its work after the 2022 election; its 290-page final report The Long Fuse remains available. | first-hand 2022-11-14 |
| ELECTION INTEGRITY PARTNERSHIP | funding insulation | 6 (ended) | independent funding — The EIP final report thanks Craig Newmark Philanthropies and the William and Flora Hewlett Foundation as the Stanford Internet Observatory's operational funders, and Knight and Hewlett as support for the UW Center for an Informed Public; no platform is named as its funder. | first-hand 2021-03-02 |
| ELECTION INTEGRITY PARTNERSHIP | reach | 6 (ended) | output acted on — The EIP final report states platforms took action on 35 percent of the URLs the Partnership reported to them (21 percent labeled, 13 percent removed, 1 percent soft blocked), with platform response rates on its Jira tickets of 76 to 90 percent. | first-hand 2021-03-02 |
| ELHAM TABASSI | durability | 4 (ended) | cited work — NIST's NFIQ 2 page lists her as NIST author and says its quality features are standardized in ISO/IEC 29794-4 as the reference implementation; the NIST AI RMF is her other cited work. | official |
| ELHAM TABASSI | insulation | 2 | at will employee — Director of the AI and Emerging Technology Initiative and Senior Fellow at Brookings; no term or protection stated. | official 2025-01-30 |
| ELHAM TABASSI | leverage | 6 (ended) | runs org acted on — NIST credits her leadership in developing the AI Risk Management Framework 1.0, a voluntary framework, and she was named CTO of the US AI Safety Institute in February 2024; she has since left NIST. | official 2023-01-26 |
| ELHAM TABASSI | leverage | 4 | advisory seat — Brookings lists her on the Policy Steering Committee of Partnership on AI, the OECD/GPAI Technical Advisory Committee and the AI Index Steering Committee. | first-hand |
| ELI TYRE | durability | 2 | publications — Project lead and main interviewer of frominside.ai, a Palisade project of interviews with frontier-lab staff. | first-hand 2026-09-29 |
| ELI TYRE | insulation | 2 | at will employee — Eli Tyre employed-by Palisade Research | official |
| ELI TYRE | leverage | 4 | advisory seat — Eli Tyre advisory Survival and Flourishing Fund | official |
| ELIEZER YUDKOWSKY | durability | 6 | built operating — Eliezer Yudkowsky founded Machine Intelligence Research Institute (MIRI) | official |
| ELIEZER YUDKOWSKY | durability | 6 | built operating — Co-founded the Machine Intelligence Research Institute in 2000 (per the page and MIRI's team page); it still operates and filed a FY2024 990. | official 2000-01-01 |
| ELIEZER YUDKOWSKY | insulation | 4 | appointed officer — MIRI's Form 990 for FY2024 lists him as Chair, Head Researcher, compensation $599,970; no control is stated. | official |
| ELIEZER YUDKOWSKY | leverage | 2 | public voice — MIRI's 23 September 2026 position on the Ban Artificial Superintelligence Act was endorsed by Bourgon, Soares and Yudkowsky on behalf of MIRI. | first-hand 2026-09-23 |
| ELIZABETH KELLY | durability | 6 (ended) | built operating — Founding director of the US AI Safety Institute, which continues as CAISI. | official 2024-02-07 |
| ELIZABETH KELLY | insulation | 2 | at will employee — Elizabeth Kelly employed-by Anthropic | official |
| ELIZABETH KELLY | leverage | 6 | runs org acted on — Head of Beneficial Deployments at Anthropic; Anthropic's Rwanda MOU announcement prints her statement in that title. | first-hand 2026-02-17 |
| ELIZABETH KELLY | leverage | 6 (ended) | runs org acted on — Commerce named her inaugural director of the US AI Safety Institute in February 2024; she departed 5 February 2025. | official 2024-02-07 |
| ELON MUSK | durability | 8 | program of record — SpaceX's S-1 states 'As of March 31, 2026, we had approximately 10.3 million Starlink Subscribers, up approximately 105% from 5.0 million subscribers a year prior'; it names Musk as founder, CEO, CTO and Chairman. | official 2026-05-20 |
| ELON MUSK | insulation | 10 | voting control — SpaceX's S-1 (filed 20 May 2026) states 'As a result, Mr. Musk will be able to control the outcome of matters requiring shareholder approval', including electing a majority of the board through Class B shares; the voting-power percentage is blank in the filed draft. | official 2026-05-20 |
| ELON MUSK | leverage | 10 | controls deployer — SpaceX's S-1 states Mr. Musk had 'a controlling financial interest in the Company, xAI and X through his majority voting interest in each such entity', and that xAI, which ships the Grok models, is now a wholly-owned subsidiary of the Company. | first-hand |
| ELON MUSK | leverage | 6 | runs org acted on — The Department of War added Starshield AI's Grok for Government to GenAI.mil on 31 August 2026, accredited for controlled unclassified information at Impact Level 5 (Air Force Personnel Center article; the page returned 403 to fetch, the same facts appear in several outlets' summaries). | official 2026-08-31 |
| EMILY OEHLSEN | durability | 1 | publications — Author of Philanthropic Cause Prioritization, Journal of Economic Perspectives 38(2), 2024. | scholarly 2024-04-01 |
| EMILY OEHLSEN | insulation | 4 | appointed officer — Coefficient Giving's team pages list her as President under CEO Alexander Berger (page unconfirmed: source returned 403). | first-hand 2025-03-06 |
| EMILY OEHLSEN | insulation | 2 | at will employee — Emily Oehlsen employed-by DeepMind | official |
| EMILY OEHLSEN | insulation | 2 | at will employee — Emily Oehlsen employed-by Open Philanthropy | official |
| EMILY OEHLSEN | leverage | 6 | runs org acted on — President of Coefficient Giving, leading its AI safety and biosecurity grantmaking since March 2025, whose recommendations Good Ventures acts on (page unconfirmed: source returned 403). | first-hand 2026-09-09 |
| EPOCH AI | entrenchment | 6 | operating — ProPublica-listed 501(c)(3) Epoch Artificial Intelligence Inc operates the benchmarking hub and publishes ongoing results (transparency page lists a Coefficient Giving general support entry of October 2026). | first-hand 2026-10-24 |
| EPOCH AI | entrenchment | 4 | cited standard — Epoch's benchmarking hub carries the banner 'Trusted by leaders at OpenAI, DeepMind, and governments worldwide' and hosts its FrontierMath benchmark results. | first-hand 2026-10-24 |
| EPOCH AI | funding insulation | 6 | independent funding — Epoch says it relies primarily on donor support and lists Coefficient Giving as its largest funder, with other listed donors; it separately lists engagements with OpenAI, Google DeepMind, xAI and Anthropic. | first-hand 2026-10-24 |
| EPOCH AI | reach | 6 | output acted on — Epoch states OpenAI commissioned Epoch AI to produce 300 advanced math problems for AI evaluation (FrontierMath), and that Epoch received OpenAI's permission before the o3 announcement to disclose the partnership. | first-hand 2025-01-23 |
| ERGO IMPACT INC | entrenchment | 6 | operating — ProPublica lists Ergo Impact Inc, EIN 99-2339731, exemption issued July 2024, with a Form 990 filed 13 May 2026 for fiscal year ending December 2025. | official 2026-05-13 |
| ERGO IMPACT INC | funding insulation | 2 | graded funding undisclosed — Ergo Impact reports TY2025 revenue of $3,731,127, all contributions, from donors not named on its site; it serves philanthropists, and its public Schedule B prints RESTRICTED for the listed contributor. | official 2026-05-13 |
| ERGO IMPACT INC | reach | 4 | convenes or funds — Ergo Impact's home page states 'We find, fund, and scale the most promising people and solutions to the world's most pressing problems' and offers philanthropists a way of deploying capital at scale. | first-hand |
| ERIC SCHMIDT | durability | 6 | built operating — He and Wendy Schmidt founded Schmidt Sciences, which still operates, and he launched SCSP in 2021, which still operates. | first-hand 2024-01-01 |
| ERIC SCHMIDT | durability | 4 | cited work — The NSCAI final report he chaired (March 2021) is archived and cited as the commission's recommendations. | official 2021-03-01 |
| ERIC SCHMIDT | insulation | 6 | independent money — SCSP's site states it is a private operating foundation and a subsidiary of The Eric & Wendy Schmidt Operating Foundation, and his bio says he launched and currently chairs it. | first-hand 2021-01-01 |
| ERIC SCHMIDT | insulation | 4 | appointed officer — Since 2022 he has served as Chairman of the Board of Directors of SandboxAQ. | first-hand 2022-01-01 |
| ERIC SCHMIDT | leverage | 6 | runs org acted on — He launched in 2021 and currently chairs the Special Competitive Studies Project, a commission-writing foundation. | first-hand 2021-01-01 |
| ERIC SCHMIDT | leverage | 4 (ended) | advisory seat — He chaired the National Security Commission on Artificial Intelligence for three years and the Defense Innovation Board for four years. | first-hand 2021-03-01 |
| ERIC SCHMIDT | leverage | 4 | funds others — Schmidt Sciences is one of the philanthropic organizations 'created and funded by Eric and Wendy Schmidt', and its trustworthy-AI programme funds academic and nonprofit safety researchers. | first-hand |
| ERIN ROBERTSON | durability | 6 | built operating — Erin Robertson founded Arcadia Impact | official |
| ERIN ROBERTSON | durability | 6 | built operating — Arcadia Impact, which she founded in 2022 and which continues to operate, reports over 50 researchers supported through LASR Labs. | first-hand 2022-01-01 |
| ERIN ROBERTSON | insulation | 2 | at will employee — Erin Robertson employed-by Open Philanthropy | official |
| ERIN ROBERTSON | insulation | 2 | at will employee — She is staff (Associate Program Officer) at Coefficient Giving, with no stated term or protection. | first-hand 2026-05-01 |
| ERIN ROBERTSON | leverage | 6 (ended) | runs org acted on — She previously founded and co-directed Arcadia Impact and was Programme Director of LASR Labs. | first-hand |
| ERIN ROBERTSON | leverage | 4 | advisory seat — She is Associate Program Officer on the Technical AI Safety team at Coefficient Giving, a philanthropic funder (recommends grants; joined May 2026). | first-hand 2026-05-01 |
| EU AI ACT | entrenchment | 10 | statute or infrastructure — Regulation (EU) 2024/1689 is an EU Regulation that entered into force on 1 August 2024 (Article 113) and applies from 2 August 2026 with staged earlier dates, so it is written into Union law. | official 2024-07-12 |
| EU AI ACT | reach | 10 | statutory power — Article 99 of Regulation (EU) 2024/1689 sets administrative fines of up to EUR 35 000 000 or 7% of total worldwide annual turnover for prohibited-practice infringements, and Article 101 lets the Commission fine general-purpose AI model providers up to 3% of worldwide turnover or EUR 15 000 000. | official 2024-07-12 |
| EU AI OFFICE | entrenchment | 10 | statute or infrastructure — Article 64(1) of Regulation (EU) 2024/1689 provides that the Commission shall develop Union expertise and capabilities in AI through the AI Office, naming the Office in the Regulation. | official 2024-07-12 |
| EU AI OFFICE | reach | 10 | statutory power — Article 101(1) of Regulation (EU) 2024/1689 lets the Commission fine providers of general-purpose AI models up to 3% of annual worldwide turnover or EUR 15 000 000; Article 3(47) defines the AI Office as the Commission's function and makes references to it references to the Commission. | official 2024-07-12 |
| EU DIGITAL SERVICES ACT | entrenchment | 10 | statute or infrastructure — Regulation (EU) 2022/2065 is directly applicable EU law that establishes the Commission's designation of very large platforms (Article 33) and its supervisory and fining powers (Articles 56 and 74). | official 2022-10-27 |
| EU DIGITAL SERVICES ACT | funding insulation | 4 | graded funding disclosed — Article 43 requires the Commission to charge each designated very large online platform and search engine an annual supervisory fee, and the overall amount of the fees must cover the estimated costs the Commission incurs for its supervisory tasks under the Regulation. | official 2022-10-27 |
| EU DIGITAL SERVICES ACT | reach | 10 | statutory power — Article 74(1) of Regulation (EU) 2022/2065 lets the Commission fine providers of very large online platforms and search engines up to 6% of total worldwide annual turnover for infringements; Article 33 has the Commission designate services with 45 million or more monthly EU users. | official 2022-10-27 |
| FABIEN ROGER | durability | 4 | cited work — Co-author of Anthropic's pilot sabotage risk report (28 Oct 2025), reviewed by METR. | first-hand 2025-10-28 |
| FABIEN ROGER | durability | 3 | cited work — Co-author of 'AI Control: Improving Safety Despite Intentional Subversion', ICML 2024 (PMLR 235), which Redwood describes as a bedrock approach to AI control. | scholarly 2024-07-01 |
| FABIEN ROGER | durability | 2 | publications — Fabien Roger co-author Holden Karnofsky | official |
| FABIEN ROGER | durability | 1 | publications — Fabien Roger co-author Ryan Greenblatt | scholarly |
| FABIEN ROGER | durability | 1 | publications — Fabien Roger co-author Buck Shlegeris | scholarly |
| FABIEN ROGER | insulation | 2 | at will employee — Fabien Roger employed-by Redwood Research | official |
| FABIEN ROGER | insulation | 2 | at will employee — Fabien Roger employed-by Anthropic | official |
| FABIEN ROGER | leverage | 4 | advisory seat — Fabien Roger advisory MATS Research | official |
| FABIEN ROGER | leverage | 2 | public voice — Fabien Roger signatory Future of Life Institute | official |
| FABIEN ROGER | leverage | 2 | public voice — He publishes AI control and alignment papers and LessWrong/Alignment Forum posts and gave a FAR.AI talk (March 2025). | first-hand 2025-03-27 |
| FACULTY | entrenchment | 6 | operating — Companies House shows Faculty Science Limited (08873131) with Accenture Plc as the active person with significant control since 12 March 2026, holding 75% or more of shares and voting rights. | official 2026-03-12 |
| FACULTY | funding insulation | 4 | parent owned — Companies House lists Accenture as the person with significant control of Faculty Science Ltd, holding 75% or more of the shares and voting rights, notified 12 March 2026. | official 2026-03-12 |
| FACULTY | reach | 6 | output acted on — Faculty evaluator-for Meta | official |
| FACULTY | reach | 6 | output acted on — Faculty evaluator-for OpenAI | official |
| FACULTY | reach | 6 | output acted on — Faculty evaluator-for DeepMind | official |
| FACULTY | reach | 6 | output acted on — Faculty evaluator-for Anthropic | official |
| FACULTY | reach | 6 | output acted on — Faculty contract UK Government | official |
| FACULTY | reach | 6 | output acted on — Faculty contract Foreign, Commonwealth & Development Office (UK) | official |
| FACULTY | reach | 6 | output acted on — Faculty contract UK AI Security Institute | official |
| FAILSPY | durability | 4 | cited work — Hugging Face lists 33 models under failspy; the models were downloaded 240,782 times in total per the platform counter, and Labonne's explainer is based on his notebook. | first-hand 2026-10-24 |
| FAILSPY | leverage | 2 | public voice — Author of the abliterator library and the abliterated Llama-3 models and model cards published under the failspy handle. | first-hand 2024-05-21 |
| FAR.AI | entrenchment | 8 | program of record — FAR.AI's EU AI Office consortium work under tender EC-CNECT/2025/OP/0032 Lot 1 runs over three years from the February 2026 announcement. | first-hand 2026-02-02 |
| FAR.AI | funding insulation | 6 | independent funding — FAR.AI's transparency page states it relies on philanthropic funding and caps revenue from for-profit AI developers at a maximum of 10% of total annual revenue. | first-hand |
| FAR.AI | reach | 8 | contract or gate — FAR.AI states it leads the CBRN Risk Modelling and Evaluation consortium under the European Commission AI Office tender EC-CNECT/2025/OP/0032 Lot 1, collaborating with the AI Office over three years. | first-hand 2026-02-02 |
| FAR.AI | reach | 6 | output acted on — FAR.AI contract European AI Office | official |
| FATHOM | entrenchment | 6 | operating — ProPublica lists Fathom AI Inc (FEIN 93-4840479) with FY2024 revenue of $12,128,490, of which contributions were $12,127,300. | official 2025-12-31 |
| FATHOM | funding insulation | 6 | independent funding — Fathom states it is a 501(c)(3) funded by philanthropists and accepts no donations from corporations, including frontier labs and FAANG companies, or from foreign entities linked to countries of concern. | first-hand 2026-08-31 |
| FATHOM | reach | 6 | output acted on — Governor Newsom signed SB 813 on 9 September 2026, establishing a framework for independent verification organizations; Fathom states it sponsored the bill and developed the IVO model at its heart. | official 2026-09-09 |
| FATHOM | reach | 4 | convenes or funds — AI Verification and Evaluation Research Institute (AVERI) funded-by Fathom AI Inc | official |
| FOUNDATION FOR AMERICAN INNOVATION | entrenchment | 6 | operating — FAI filed a Form 990 for 2024 reporting total revenue of $5,149,927, of which contributions were $5,106,107. | official 2025-11-12 |
| FOUNDATION FOR AMERICAN INNOVATION | reach | 6 | output acted on — FAI's Q2 2026 update says an executive order reflects the case its fellow Dean Ball made for voluntary pre-deployment evaluation of frontier models, and that Commerce's CAISI signed evaluation agreements with Google DeepMind, Microsoft and xAI. | first-hand 2026-07-01 |
| FOUNDATION FOR AMERICAN INNOVATION | reach | 4 | convenes or funds — FAI runs the Conservative AI Policy Fellowship, an eight-week part-time fellowship for conservative policy professionals in Washington, DC. | first-hand |
| FRONTIER DESIGN | entrenchment | 6 | operating — Frontier Design Group LLC's site lists Alexa Courtney as Founder + CEO and carries a 2026 copyright notice for Frontier Design Group LLC. | first-hand 2026-10-23 |
| FRONTIER DESIGN | reach | 6 | output acted on — Frontier Design evaluator-for Meta | official |
| FRONTIER DESIGN | reach | 6 | output acted on — Frontier Design evaluator-for Anthropic | official |
| FRONTIER DESIGN | reach | 6 | output acted on — Anthropic's Fable 5 and Mythos 5 system card names the Frontier Design beneficial red-teaming tabletop exercise and says it produced the strongest CB-2 signal of any single evaluation. | official 2026-06-09 |
| FRONTIER DESIGN | reach | 6 | output acted on — Frontier Design contract Frontier Model Forum AI Safety Fund | official |
| FRONTIER MODEL FORUM | entrenchment | 6 | operating — The Forum's membership page states 'The current members of the FMF are Amazon, Anthropic, Google, Meta, Microsoft, and OpenAI.' | first-hand |
| FRONTIER MODEL FORUM | funding insulation | 4 | graded funding disclosed — The Forum's membership page names its members (Amazon, Anthropic, Google, Meta, Microsoft, OpenAI) and says members commit to 'financially supporting the FMF's safety efforts for a minimum of three years'. | first-hand |
| FRONTIER MODEL FORUM | reach | 6 | output acted on — xAI's Frontier AI Framework states 'xAI references standards such as NIST's AI Risk Management Framework, ISO/IEC 42001 ... and industry best practices from the Frontier Model Forum (e.g., red-teaming protocols).' | official 2025-12-31 |
| FUTURE OF LIFE INSTITUTE | entrenchment | 6 | operating — FLI's finances page reports 2025 expenditure and says it engaged in limited fundraising in 2025 and covered the gap from its endowment-like funds, so it operates today. | first-hand 2026-10-23 |
| FUTURE OF LIFE INSTITUTE | funding insulation | 8 | endowment or appropriation — FLI states that Vitalik Buterin's 2021 unconditional gift, part of which serves as an endowment, provides long-term stability, and that its 2025 shortfall was covered by funds functioning as an endowment. | first-hand 2026-10-23 |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Collective Intelligence Project funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — FLI's finances page reports 2025 expenditure that includes $26,724,000 of grants to charitable organizations. | first-hand 2026-10-23 |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Center for Humane Technology funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Centre for the Study of Existential Risk (Cambridge) funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Jeffrey Ding funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — LawZero funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Future of Life Foundation (FLF) funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Good Ventures funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Lightcone Foundation (Nevada) funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Berkeley Existential Risk Initiative (BERI) funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Longview Philanthropy funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Survival and Flourishing Fund funded-by Future of Life Institute | official |
| FUTURE OF LIFE INSTITUTE | reach | 4 | convenes or funds — Miles Brundage funded-by Future of Life Institute | official |
| GARM + GIFCT | entrenchment | 6 | operating — Global Internet Forum To Counter Terrorism Inc (EIN 84-4586487) reported 2024 revenue of $3,207,102 on its Form 990. | official 2024-12-31 |
| GARM + GIFCT | funding insulation | 4 | graded funding disclosed — GIFCT's membership page states that all members are required to contribute annual membership dues based on their membership tier. | first-hand |
| GARM + GIFCT | funding insulation | 0 (ended) | defunded or shut — The World Federation of Advertisers announced that GARM, its voluntary cross-industry initiative created in 2019, will discontinue its activities. | first-hand 2024-08-09 |
| GARM + GIFCT | reach | 6 | output acted on — GIFCT's Hash-Sharing Database lets member companies share signals of terrorist and violent extremist content; GIFCT says each member decides what action to take on matching content. | first-hand |
| GATES FOUNDATION | entrenchment | 6 | operating — The Gates Foundation publishes audited financial statements and Form 990-PF filings and operates through the Foundation and the Gates Foundation Trust. | first-hand |
| GATES FOUNDATION | funding insulation | 8 | endowment or appropriation — The foundation says its trust holds the donated investment assets of Bill Gates, Melinda French Gates and Warren Buffett and transfers proceeds to the foundation as needed for its charitable goals. | first-hand |
| GATES FOUNDATION | reach | 4 | convenes or funds — WHO's programme budget portal lists the Gates Foundation as a 2024-25 contributor with total voluntary contributions of 761,065K US dollars, the top-listed donor ahead of the United States. | official 2026-10-24 |
| GATES FOUNDATION | reach | 4 | convenes or funds — Atlantic Council funded-by Bill & Melinda Gates Foundation | official |
| GATES FOUNDATION | reach | 4 | convenes or funds — eGovernments Foundation funded-by Bill & Melinda Gates Foundation | official |
| GATES FOUNDATION | reach | 4 | convenes or funds — Global Impact funded-by Bill & Melinda Gates Foundation | official |
| GATES FOUNDATION | reach | 4 | convenes or funds — Institute for Strategic Dialogue funded-by Bill & Melinda Gates Foundation | official |
| GATES FOUNDATION | reach | 4 | convenes or funds — World Bank funded-by Bill & Melinda Gates Foundation | official |
| GATES FOUNDATION | reach | 4 | convenes or funds — Nesta funded-by Bill & Melinda Gates Foundation | official |
| GATES FOUNDATION | reach | 4 | convenes or funds — LawZero funded-by Bill & Melinda Gates Foundation | official |
| GENERALITY LABS | entrenchment | 6 | operating — The Inspect Evals README states the library is maintained by Generality Labs, a London-based nonprofit, and Generality says it was founded in July 2026 with $4.4M seed funding from Coefficient Giving. | first-hand 2026-10-14 |
| GENERALITY LABS | reach | 8 (ended) | contract or gate — The same proposal states that ASET (Arcadia Impact) has been contracted with the UK AISI since January 2025 to manage Inspect Evals. | first-hand 2025-01-01 |
| GENERALITY LABS | reach | 6 | output acted on — Generality Labs contract UK AI Security Institute | official |
| GENERALITY LABS | reach | 6 | output acted on — Justin Olive's Manifund proposal for Inspect Evals states it is used by the UK AISI, METR, RAND, US CAISI and other organisations for safety testing and capability assessment. | first-hand 2025-09-01 |
| GEOFF RALSTON | durability | 6 | built operating — Geoff Ralston founded SAIF (Safe Artificial Intelligence Fund) | official |
| GEOFF RALSTON | durability | 6 | built operating — Geoff Ralston founded SAIFbio Inc. | official |
| GEOFF RALSTON | insulation | 4 | appointed officer — He serves as President of SAIFbio Inc., a 501(c)(3) operated by the partners of SAIF Ventures. | first-hand 2026-06-01 |
| GEOFF RALSTON | insulation | 2 | at will employee — Geoff Ralston employed-by Y Combinator | official |
| GEOFF RALSTON | leverage | 4 | advisory seat — Geoff Ralston advisory MATS Research | official |
| GEOFF RALSTON | leverage | 4 (ended) | advisory seat — Former President of Y Combinator (early 2019 to end 2022). | first-hand 2022-12-31 |
| GEOFF RALSTON | leverage | 4 | funds others — METR (Model Evaluation & Threat Research) funded-by Geoff Ralston | official |
| GEOFF RALSTON | leverage | 4 | funds others — He founded SAIF (Safe Artificial Intelligence Fund), which writes $100K investments into AI-safety startups (about 30 funded in year one by his account). | first-hand 2025-04-17 |
| GEOFF RALSTON | leverage | 4 | funds others — AI Verification and Evaluation Research Institute (AVERI) funded-by Geoff Ralston | official |
| GEOFFREY HINTON | durability | 6 | built operating — Geoffrey Hinton founded Vector Institute for Artificial Intelligence | official |
| GEOFFREY HINTON | durability | 4 | cited work — The Nobel Prize in Physics 2024 was awarded to Hopfield and Hinton 'for foundational discoveries and inventions that enable machine learning with artificial neural networks'. | official 2024-10-08 |
| GEOFFREY HINTON | insulation | 6 | independent money — His own University of Toronto page: 'he is now an emeritus distinguished professor'. Emeritus is a retained title with no salary or term stated, so tenure is the nearest type, not a proven protected term. | first-hand |
| GEOFFREY HINTON | insulation | 2 | at will employee — Geoffrey Hinton employed-by Google / Alphabet | official |
| GEOFFREY HINTON | leverage | 4 | advisory seat — Vector Institute calls him 'Co-founder and Chief Scientific Advisor of Vector'. | first-hand 2024-10-08 |
| GEOFFREY HINTON | leverage | 2 | public voice — Geoffrey Hinton signatory Center for AI Safety | official |
| GEOFFREY HINTON | leverage | 2 | public voice — Signed the CAIS Statement on AI Risk (May 2023) and the FLI Statement on Superintelligence (22 Oct 2025); Nobel banquet speech 10 Dec 2024: 'We have no idea whether we can stay in control'. | first-hand 2024-12-10 |
| GEOFFREY IRVING | durability | 6 | built operating — Resolution (formerly Timaeus Research), which he co-founded in June 2026, has a $160M grant from Coefficient Giving. | first-hand 2026-07-07 |
| GEOFFREY IRVING | durability | 6 | built operating — Geoffrey Irving founded Resolution | official |
| GEOFFREY IRVING | durability | 3 | cited work — Lead author of 'AI safety via debate' (arXiv 1805.00899, 2018) and co-author of AISI's debate and control safety-case sketches. | scholarly 2018-05-02 |
| GEOFFREY IRVING | insulation | 4 | appointed officer — He is co-founder and Chief Scientist of Resolution, whose launch post says its directors 'report up to Geoffrey Irving'; its sole funder to start is Coefficient Giving. | first-hand 2026-06-10 |
| GEOFFREY IRVING | insulation | 2 | at will employee — Geoffrey Irving employed-by DeepMind | official |
| GEOFFREY IRVING | insulation | 2 | at will employee — Geoffrey Irving employed-by OpenAI | official |
| GEOFFREY IRVING | insulation | 2 | at will employee — Geoffrey Irving employed-by UK AI Security Institute | official |
| GEOFFREY IRVING | leverage | 6 (ended) | runs org acted on — He was Research Director and then Chief Scientist at the UK AI Security Institute, which tests frontier models. | first-hand 2026-05-15 |
| GEOFFREY IRVING | leverage | 4 | advisory seat — Geoffrey Irving advisory LawZero | official |
| GEOFFREY IRVING | leverage | 4 | advisory seat — He says he will continue 'in an advisory capacity at AISI' after leaving, and LawZero lists him a Scientific Advisor. | first-hand 2026-05-15 |
| GILLIAN HADFIELD | durability | 6 (ended) | built operating — She was inaugural director of the Schwartz Reisman Institute, University of Toronto (2019-2023), which continues to operate. | first-hand 2019-01-01 |
| GILLIAN HADFIELD | durability | 4 | cited work — Her regulatory-markets proposal (Hadfield and Clark, Jurimetrics vol. 65, 2026) is the model her site says helped shape the IVO framework introduced in state and federal legislation; California SB 813 (Ch. 179, 2026) creates designated independent verification organizations. | official 2026-01-01 |
| GILLIAN HADFIELD | durability | 1 | publications — Gillian Hadfield co-author Logan Graham | scholarly |
| GILLIAN HADFIELD | durability | 1 | publications — Gillian Hadfield co-author Miles Brundage | scholarly |
| GILLIAN HADFIELD | durability | 1 | publications — Gillian Hadfield co-author Jack Clark | scholarly |
| GILLIAN HADFIELD | insulation | 4 | appointed officer — She serves as Board Chair of the Cooperative AI Foundation. | first-hand 2025-01-01 |
| GILLIAN HADFIELD | insulation | 2 | at will employee — Gillian Hadfield employed-by OpenAI | official |
| GILLIAN HADFIELD | leverage | 4 | advisory seat — Gillian Hadfield advisory Fathom AI Inc | official |
| GILLIAN HADFIELD | leverage | 4 | advisory seat — Gillian Hadfield advisory RAND | official |
| GILLIAN HADFIELD | leverage | 4 | advisory seat — Gillian Hadfield advisory AI Verification and Evaluation Research Institute (AVERI) | official |
| GILLIAN HADFIELD | leverage | 4 | advisory seat — Named 23 Sept 2026 to Governor Newsom's group of experts advising on recommendations under his AI executive order. | official 2026-09-23 |
| GILLIAN HADFIELD | leverage | 4 | advisory seat — Member of Canada's National Council on Artificial Intelligence (launched 2 Oct 2026). | official 2026-10-02 |
| GLOBAL DISINFORMATION INDEX | entrenchment | 6 | operating — Disinformation Index Inc filed a Form 990 for calendar 2024 reporting total revenue of $3,006,917, including $1,491,846 of research licensing revenue. | official 2025-10-29 |
| GLOBAL DISINFORMATION INDEX | funding insulation | 6 | independent funding — GDI's About page states under Independence: We accept no funding from media companies or technology platforms. | first-hand |
| GLOBAL DISINFORMATION INDEX | reach | 6 (ended) | output acted on — GDI's amended complaint against the FTC states each subscriber was free to use its dynamic exclusion list to inform independent brand safety decisions, and that GDI has since stopped publishing the list. | first-hand 2026-02-02 |
| GLOBAL DISINFORMATION INDEX | reach | 4 | convenes or funds — GDI's Our Work page states that GDI has advised Ofcom on the Online Safety Act and is a signatory to the EU Code of Conduct on Disinformation. | first-hand |
| GLOBAL ENGAGEMENT CENTER | funding insulation | 0 | defunded or shut — The CRS states the GEC's legislative authority terminated on 23 December 2024 after Congress declined to enact several proposals to extend it, and the State Department confirmed it closed. | official 2024-12-26 |
| GLOBAL ENGAGEMENT CENTER | reach | 4 (ended) | convenes or funds — The State Department Inspector General recorded that GEC federal assistance programs totaled approximately 15.7 million dollars in FY 2019-2020 and 16.9 million in FY 2020-2021. | official 2022-09 |
| GOOD VENTURES | entrenchment | 6 | operating — Good Ventures Foundation (EIN 46-1008520, IRS ruling March 2013) filed a Form 990-PF on 15 May 2026 for the year to June 2025. | official 2026-05-15 |
| GOOD VENTURES | funding insulation | 8 | endowment or appropriation — Good Ventures Foundation reported total assets of $10,107,955,038 and total revenue of $2,597,571,672 for the year to June 2025, a private foundation endowed by its founders. | official 2026-05-15 |
| GOOD VENTURES | reach | 4 | convenes or funds — Open Philanthropy funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — Jeffrey Ding funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — Epoch AI funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — FAR.AI funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — Centre for the Governance of AI funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — Irregular (formerly Pattern Labs) funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — MATS Research funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — RAND funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — SecureBio funded-by Good Ventures | official |
| GOOD VENTURES | reach | 4 | convenes or funds — Good Ventures Foundation's latest return (year to June 2025) shows charitable disbursements of $337,695,777 and total assets of $10,107,955,038. | official 2026-05-15 |
| GOOGLE DEEPMIND | entrenchment | 8 | program of record — Alphabet states the Gemini app now has 950 million monthly active users. | first-hand 2026-07-22 |
| GOOGLE DEEPMIND | entrenchment | 8 | program of record — The UK DSIT memorandum with Google DeepMind will remain active for 36 months and gives the UK AI Security Institute priority technical access to its frontier models. | official 2025-12-11 |
| GOOGLE DEEPMIND | funding insulation | 4 | parent owned — Companies House lists DeepMind Holdings Limited as the person with significant control of DeepMind Technologies Limited, with ownership of shares and voting rights of 75% or more and the right to appoint or remove directors; Alphabet, Inc. held the same control until 4 November 2019. | official 2019-11-04 |
| GOOGLE DEEPMIND | reach | 8 | contract or gate — GSA announced a OneGov agreement with Google for Gemini for Government, with agencies paying $0.47 per agency, valid through 2026. | official 2025-08-21 |
| GOOGLE DEEPMIND | reach | 4 | convenes or funds — AI Verification and Evaluation Research Institute (AVERI) funded-by DeepMind | official |
| GOVAI (CENTRE FOR THE GOVERNANCE OF AI) | entrenchment | 6 | operating — GovAI's 2025 annual report describes an operating organization in its first full year as an independent nonprofit, with 11.9 million dollars of 2025 expenses. | first-hand |
| GOVAI (CENTRE FOR THE GOVERNANCE OF AI) | funding insulation | 6 | independent funding — GovAI's 2025 annual report names its funders as Coefficient Giving, the Survival and Flourishing Fund, Longview Philanthropy and donors on Giving What We Can and Every.org, with no AI developer named, against 11.9 million dollars of 2025 expenses. | first-hand |
| GOVAI (CENTRE FOR THE GOVERNANCE OF AI) | reach | 4 | convenes or funds — GovAI's 2025 annual report records its hosting of closed events and roundtables and Markus Anderljung serving as vice-chair in drawing up the EU Code of Practice for general-purpose AI with systemic risk. | first-hand |
| GRACE WERNER | durability | 6 | built operating — She co-founded AVERI, which continues to operate. | first-hand 2025-01-01 |
| GRACE WERNER | durability | 4 | cited work — One of 26 authors of the AI Evaluator Forum's AEF-1 minimum operating conditions standard (v1, 4 Dec 2025). | first-hand 2025-12-04 |
| GRACE WERNER | insulation | 4 | appointed officer — Vice President of Growth at the Foundation for American Innovation (Q1 2026). | first-hand 2026-01-01 |
| GRACE WERNER | insulation | 2 | at will employee — Grace Werner employed-by Foundation for American Innovation | official |
| GRACE WERNER | leverage | 4 | advisory seat — Grace Werner board-seat AI Verification and Evaluation Research Institute (AVERI) | official |
| GRACE WERNER | leverage | 4 | convenes — Co-organizer of the DC AI Security Forum (National Press Club, 9 July 2025). | first-hand 2025-07-09 |
| GRAPHIKA | entrenchment | 8 | program of record — USAspending lists 14 Department of the Air Force contracts to Octant Data LLC dba Graphika Technologies totalling $8,651,891, including FA714624C0014 ($1,899,332, to 24 March 2026) and FA714625C0022 (to 19 November 2026). | official 2024-06-18 |
| GRAPHIKA | reach | 8 | contract or gate — USAspending lists an Air Force definitive contract FA714625C0022 of $1,995,359.78 to Octant Data LLC (Graphika), running 19 May 2025 to 19 Nov 2026. | official 2025-05-19 |
| GRAPHIKA | reach | 6 | output acted on — Graphika contract DoD | official |
| GRAY SWAN | entrenchment | 6 | operating — Gray Swan's release states it works with over 20 customers across frontier labs and global enterprises and is cited in 11 recent frontier model system cards. | first-hand 2026-05-28 |
| GRAY SWAN | funding insulation | 6 | independent funding — Gray Swan's release states a $40 million Series A co-led by Wing Venture Capital and Madrona, with Obvious Ventures, Snowflake Ventures, Hudson River Trading, Samsung Next and Magarac Venture Partners. | first-hand 2026-05-28 |
| GRAY SWAN | reach | 8 (ended) | contract or gate — UK Contracts Finder lists a 128,770 GBP award for AI Research - Robust Safeguards and Offensive Cyber Capability Measurement provided by GraySwan AI. | official 2024-01-15 |
| GRAY SWAN | reach | 6 | output acted on — Gray Swan AI evaluator-for Meta | official |
| GRAY SWAN | reach | 6 | output acted on — Gray Swan AI evaluator-for OpenAI | official |
| GRAY SWAN | reach | 6 | output acted on — Gray Swan AI evaluator-for DeepMind | official |
| GRAY SWAN | reach | 6 | output acted on — Gray Swan AI evaluator-for Anthropic | official |
| GRAY SWAN | reach | 6 | output acted on — Gray Swan AI contract UK AI Security Institute | official |
| GREG BROCKMAN | durability | 6 | built operating — He co-founded OpenAI in December 2015 as founding CTO; it still operates. | first-hand 2015-12-11 |
| GREG BROCKMAN | insulation | 8 | reinstated — OpenAI's page on the reinstatement under a new initial board says 'Greg Brockman returns as President'; the new board concluded 8 Mar 2024 that 'Sam and Greg are the right leaders'. | first-hand 2023-11-29 |
| GREG BROCKMAN | insulation | 0 (reversed) | removed not reinstated — OpenAI announced 17 Nov 2023 that he would step down as chairman of the board. | first-hand 2023-11-17 |
| GREG BROCKMAN | leverage | 5 | runs org acted on — President and co-founder of OpenAI (President since 5 May 2022); signed the White House frontier accord for OpenAI on 29 Sept 2026. | press 2026-09-29 |
| GREG BROCKMAN | leverage | 4 | funds others — He and Anna Brockman each gave $12,500,000 to the Leading the Future super PAC on 12 Sept 2025, and he gave $12,500,000 to MAGA Inc. | official 2025-09-12 |
| GREG BROCKMAN | leverage | 4 | funds others — Leading the Future (super PAC, FEC C00916114) funded-by Greg Brockman | official |
| GRYPHON SCIENTIFIC | entrenchment | 2 (ended) | reports persist — NIH OSP hosts Gryphon's Risk and Benefit Analysis of Gain of Function Research, prepared under NIH Contract HHSN263201500002C, 20 March to 15 December 2015. | official 2015-12-15 |
| GRYPHON SCIENTIFIC | reach | 8 (ended) | contract or gate — UK DSIT awarded Gryphon Scientific LLC a 267,618.41 GBP contract, Analysis and Research to support the UK AISI's AI Safety Programmes, without competition because the services can only be supplied by Gryphon. | official 2024-01-23 |
| GRYPHON SCIENTIFIC | reach | 6 | output acted on — OpenAI's GPT-6 Astra card grades tacit knowledge and troubleshooting with a multiple-choice dataset built in-house with its partners at Gryphon Scientific, not published. | official 2026-10-01 |
| GRYPHON SCIENTIFIC | reach | 6 | output acted on — Gryphon Scientific contract National Institutes of Health | official |
| GRYPHON SCIENTIFIC | reach | 6 | output acted on — Gryphon Scientific contract UK AI Security Institute | official |
| HALCYON FUTURES | entrenchment | 6 | operating — Halcyon Futures Inc, a Delaware 501(c)(3) with tax exemption issued April 2024, filed a Form 990 for calendar 2024 reporting revenue of $2,116,383. | official 2025-11-17 |
| HALCYON FUTURES | reach | 4 | convenes or funds — Halcyon Futures Inc reports $611,455 in grants and similar amounts paid in calendar 2024 to cohort members and other nonprofits, per its Form 990. | official 2025-11-17 |
| HALCYON FUTURES | reach | 4 | convenes or funds — Seismic Catalyst Foundation funded-by Halcyon Futures | official |
| HALCYON FUTURES | reach | 4 | convenes or funds — AI Verification and Evaluation Research Institute (AVERI) funded-by Halcyon Futures | official |
| HALCYON FUTURES | reach | 4 | convenes or funds — Artificial Intelligence Underwriting Company (AIUC) funded-by Halcyon Futures | official |
| HALCYON FUTURES | reach | 4 | convenes or funds — Transluce funded-by Halcyon Futures | official |
| HAMILTON 68 / ALLIANCE FOR SECURING DEMOCRACY | entrenchment | 6 (ended) | operating — ASD ran the Hamilton 68 dashboard from August 2017 until it discontinued it in December 2018. | first-hand 2018-12-31 |
| HAMILTON 68 / ALLIANCE FOR SECURING DEMOCRACY | entrenchment | 2 | reports persist — ASD's methodology note and fact sheet on the Hamilton 68 dashboard remain published; the fact sheet states ASD discontinued the dashboard in December 2018. | first-hand 2023-01-27 |
| HAMILTON 68 / ALLIANCE FOR SECURING DEMOCRACY | funding insulation | 6 (ended) | independent funding — ASD states it receives funding primarily from philanthropic foundations and individual donors and no financial support from social media companies. | first-hand |
| HAMILTON 68 / ALLIANCE FOR SECURING DEMOCRACY | reach | 2 (ended) | publishes — ASD launched the Hamilton 68 dashboard in August 2017 and published its methodology on 7 August 2017, stating it monitored 600 Twitter accounts linked to Russian influence efforts and chose not to identify them. | first-hand 2017-08-07 |
| HANDSHAKE AI | entrenchment | 6 | operating — Stryder Corp. d/b/a Handshake operates Handshake AI, a human-data and red-team business, and is a named defendant in Boggs v. Stryder Corp., N.D. Cal. 3:26-cv-07498, filed 20 July 2026. | official 2026-07-20 |
| HANDSHAKE AI | reach | 6 | output acted on — Handshake AI evaluator-for Meta | official |
| HELEN TONER | durability | 6 | built operating — Georgetown's dean says she has been 'instrumental in building CSET from the ground up' since its 2019 founding; CSET still operates. | first-hand 2019-01-01 |
| HELEN TONER | durability | 4 | cited work — Co-author of CSET's October 2023 report 'Decoding Intentions: Artificial Intelligence and Costly Signals'. | first-hand 2023-10-01 |
| HELEN TONER | durability | 1 | publications — Helen Toner co-author Miles Brundage | scholarly |
| HELEN TONER | insulation | 4 | appointed officer — CSET 'has appointed Helen Toner as Interim Executive Director', succeeding Dewey Murdick. | first-hand 2025-09-02 |
| HELEN TONER | insulation | 2 | at will employee — Helen Toner employed-by Open Philanthropy | official |
| HELEN TONER | insulation | 0 (ended) | removed not reinstated — She was one of four OpenAI directors who removed Altman on 17 Nov 2023, and left the board in the 29 Nov 2023 settlement; she was not reinstated. | first-hand 2023-11-29 |
| HELEN TONER | leverage | 6 | runs org acted on — Interim Executive Director of Georgetown's CSET effective 2 Sept 2025. | first-hand 2025-09-02 |
| HELEN TONER | leverage | 4 | advisory seat — Testified to the Senate Judiciary Committee (22 Apr 2026) and its subcommittee (17 Sept 2024); GovAI board secretary per its Form 990. | official 2025-11-13 |
| HELEN TONER | leverage | 4 | advisory seat — Helen Toner board-seat Centre for the Governance of AI | official |
| HENRY KISSINGER | durability | 2 (ended) | publications — Co-author of The Age of AI (Little, Brown, 2 Nov 2021) and Genesis (19 Nov 2024); author of 'How the Enlightenment Ends' (The Atlantic, June 2018). | first-hand 2021-11-02 |
| HENRY KISSINGER | insulation | 0 (ended) | removed not reinstated — He was among eleven advisers removed from the Defense Policy Board on 25 Nov 2020; no reinstatement shown. | press 2020-11-25 |
| HENRY KISSINGER | leverage | 3 (ended) | advisory seat — Member of the Defense Policy Board until removed by a November 2020 directive. | press 2020-11-25 |
| WILLIAM AND FLORA HEWLETT FOUNDATION | entrenchment | 6 | operating — The foundation filed a Form 990-PF for fiscal year 2024 on 14 November 2025 and committed $100 million through 2031 in July 2026. | official 2025-11-14 |
| WILLIAM AND FLORA HEWLETT FOUNDATION | funding insulation | 8 | endowment or appropriation — The foundation's Form 990-PF for 2024 reports total assets of $14,202,751,280, revenue of $784,700,540 and charitable disbursements of $689,557,086. | official 2025-11-14 |
| WILLIAM AND FLORA HEWLETT FOUNDATION | reach | 4 | convenes or funds — Carnegie Endowment for International Peace funded-by William and Flora Hewlett Foundation | official |
| WILLIAM AND FLORA HEWLETT FOUNDATION | reach | 4 | convenes or funds — The Hewlett Foundation announced on 16 July 2026 a $100 million, five-year Emerging Technology and Security Initiative to fund civil society groups, universities and think tanks on AI, biotechnology and quantum security. | first-hand 2026-07-16 |
| WILLIAM AND FLORA HEWLETT FOUNDATION | reach | 4 | convenes or funds — Atlantic Council funded-by William and Flora Hewlett Foundation | official |
| WILLIAM AND FLORA HEWLETT FOUNDATION | reach | 3 | convenes or funds — Stanford Internet Observatory funded-by William and Flora Hewlett Foundation | scholarly |
| HIDDENLAYER | entrenchment | 6 | operating — HiddenLayer announced on 23 December 2025 it was selected as an awardee on the Missile Defense Agency SHIELD IDIQ vehicle with a $151 billion ceiling; no task order to HiddenLayer is named. | first-hand 2025-12-23 |
| HIDDENLAYER | reach | 8 | contract or gate — HiddenLayer announced on 9 April 2024 that AFWERX selected it for an STTR Phase II contract of $1.8 million focused on AI Detection and Response for the Department of the Air Force. | first-hand 2024-04-09 |
| HJALMAR WIJK | durability | 3 | cited work — Wijk is first author of RE-Bench (arXiv 2411.15114), a benchmark of 7 research-engineering environments with 71 human expert attempts. | scholarly 2024-11-22 |
| HJALMAR WIJK | durability | 2 | publications — Hjalmar Wijk co-author Ajeya Cotra | official |
| HJALMAR WIJK | durability | 2 | publications — Hjalmar Wijk co-author Ryan Greenblatt | official |
| HJALMAR WIJK | durability | 1 | publications — Hjalmar Wijk co-author Paul Christiano | scholarly |
| HJALMAR WIJK | durability | 1 | publications — Hjalmar Wijk co-author Beth Barnes | scholarly |
| HJALMAR WIJK | durability | 1 | publications — Hjalmar Wijk co-author Chris Painter | scholarly |
| HJALMAR WIJK | insulation | 2 | at will employee — Hjalmar Wijk employed-by METR (Model Evaluation & Threat Research) | official |
| HJALMAR WIJK | insulation | 2 | at will employee — Hjalmar Wijk employed-by Alignment Research Center | official |
| HJALMAR WIJK | leverage | 6 | runs org acted on — Hjalmar Wijk is Chief Scientist of METR, whose Frontier Risk Report assessed models of Anthropic, Google, Meta and OpenAI with company engagement. | first-hand |
| HJALMAR WIJK | leverage | 4 | advisory seat — Hjalmar Wijk advisory European Commission | official |
| HOAN TON-THAT | durability | 8 (ended) | program of record — USAspending lists multi-year ICE contract awards to Clearview AI that began in his tenure, e.g. ICE 70CMSD21P00000127 ($2,287,500, 3 Sep 2021 to 3 Sep 2024). | official 2021-09-03 |
| HOAN TON-THAT | durability | 5 (ended) | built operating — Ton-That co-founded Clearview AI in 2017 and as executive officer and director signed its Form D for an $8,625,000 raise; the company continued operating under successor co-CEOs Lambert and Schwartz. | press 2025-02-21 |
| HOAN TON-THAT | insulation | 2 | at will employee — Hoan Ton-That employed-by Architect (San Francisco asset manager) | official |
| HOAN TON-THAT | insulation | 0 (ended) | removed not reinstated — Shareholders voted Ton-That off the Clearview AI board in April 2025; he told The Register 'I was unexpectedly removed from the Clearview AI board'. | press 2025-04-08 |
| HOAN TON-THAT | leverage | 6 (ended) | runs org acted on — Ton-That signed Clearview AI's Form D as Chief Executive Officer and is listed as executive officer and director; Clearview's face-search product was licensed to ICE, FBI and other agencies. | official 2020-09-24 |
| HOLDEN KARNOFSKY | durability | 6 (ended) | built operating — Karnofsky and Elie Hassenfeld started GiveWell full-time in 2007; GiveWell incubated the Open Philanthropy Project, which became a separate entity in June 2017. | first-hand 2007-01-01 |
| HOLDEN KARNOFSKY | insulation | 1 | at will employee — Anthropic hired Karnofsky as a member of the technical staff reporting to Jared Kaplan, working on the responsible scaling policy; no term or protection stated. | press 2025-02-13 |
| HOLDEN KARNOFSKY | leverage | 8 (ended) | signs grants — Karnofsky left Open Philanthropy for Carnegie saying his work 'no longer involves significant involvement in grantmaking', having previously overseen its grantmaking as co-CEO. | first-hand 2024-04-29 |
| HOLDEN KARNOFSKY | leverage | 4 | advisory seat — Holden Karnofsky board-seat Alignment Research Center | official |
| HOLDEN KARNOFSKY | leverage | 2 | public voice — Karnofsky publishes Cold Takes and gave the 2025 80,000 Hours podcast interview on AI safety at frontier companies. | first-hand 2025-10-30 |
| HUGGING FACE | entrenchment | 8 | program of record — Hugging Face reports 13 million users, more than 2 million public models and over 500,000 public datasets in 2025, and that over 30 percent of the Fortune 500 maintain verified accounts. | first-hand 2026-03-17 |
| HUGGING FACE | reach | 2 | publishes — Hugging Face publishes its State of Open Source: Spring 2026 report and the Hub's model, dataset and evaluation pages. | first-hand 2026-03-17 |
| HUMANE INTELLIGENCE | entrenchment | 6 | operating — Humane Intelligence Inc, a Texas 501(c)(3) with exemption issued March 2024, filed Form 990 returns for 2024 and 2025 reporting evaluation and red-teaming program revenue. | official 2026-08-26 |
| HUMANE INTELLIGENCE | funding insulation | 2 (ended) | graded funding undisclosed — Humane Intelligence Inc's Form 990 for 2024 reports revenue of $472,206, of which $372,175 is program-service revenue from AI evaluation work and $100,000 is contributions; the paying clients are not named. | official 2026-08-26 |
| HUMANE INTELLIGENCE | reach | 4 | convenes or funds — Singapore's AI Safety Institute page states IMDA, together with Humane Intelligence, completed an AI Safety Red Teaming Challenge held 5 November 2024 with over 350 participants across 9 countries testing 4 large language models. | official 2024-11-05 |
| IAN HOGARTH | durability | 6 | built operating — The register describes the AI Safety Institute as 'formerly the AI Foundation Model Taskforce', which Hogarth was appointed to chair on 12 June 2023; the Institute still operates as the AI Security Institute. | official 2023-06-12 |
| IAN HOGARTH | insulation | 4 | appointed officer — Hogarth holds the Chair by appointment under an agreement with DSIT's Permanent Secretary; the register's last appointment ran to 9 December 2025 with no reappointment notice found. | official 2025-02-10 |
| IAN HOGARTH | leverage | 6 | runs org acted on — The government register lists Hogarth as Chair of the AI Security Institute (formerly AI Safety Institute and Foundation Model Taskforce), 12 June 2023 to 9 December 2025; the Institute's About page still lists 'Our Chair Ian Hogarth'. | official 2023-06-12 |
| IAN HOGARTH | leverage | 4 | funds others — Hogarth is General Partner of Plural, a venture fund; Plural's team page lists his investments (Proxima Fusion, Hypersonica, Phasecraft, Callosum). | first-hand 2026-05-28 |
| ILYA SUTSKEVER | durability | 6 (ended) | built operating — OpenAI's founding post names Sutskever as research director; OpenAI continues to operate after his May 2024 departure. | first-hand 2015-12-11 |
| ILYA SUTSKEVER | durability | 6 | built operating — Sutskever founded Safe Superintelligence Inc., which states it is an American company with offices in Palo Alto and Tel Aviv pursuing one product, a safe superintelligence; he states he is formally CEO (July 2025). | first-hand 2024-06-19 |
| ILYA SUTSKEVER | durability | 6 | built operating — Ilya Sutskever co-founder Safe Superintelligence Inc. | official |
| ILYA SUTSKEVER | leverage | 6 (ended) | runs org acted on — OpenAI announced its Superalignment team 'co-led by Ilya Sutskever and Jan Leike' with 20% of secured compute; he left OpenAI in May 2024. | first-hand 2023-07-05 |
| ILYA SUTSKEVER | leverage | 2 | public voice — Sutskever gave a long Dwarkesh Podcast interview (25 Nov 2025) on SSI's funding, compute and Meta's acquisition offer. | first-hand 2025-11-25 |
| INTERNATIONAL MONETARY FUND | entrenchment | 10 | statute or infrastructure — The IMF Articles of Agreement, Article XXVIII, require acceptance by three-fifths of members holding eighty-five percent of total voting power to amend the charter, and Article IV Section 3 obliges the Fund to exercise firm surveillance over members' exchange rate policies. | official |
| INTERNATIONAL MONETARY FUND | funding insulation | 8 | endowment or appropriation — The IMF's Annual Report 2026 states IMF quotas total SDR 476 billion (about $653 billion) and that it generates income primarily through lending and investing activities. | first-hand 2026-04-30 |
| INTERNATIONAL MONETARY FUND | reach | 8 | contract or gate — On 26 February 2025 the IMF Executive Board approved a 40-month arrangement for El Salvador with access equivalent to US$1.4 billion, listing as a prior action legal reforms making private-sector acceptance of Bitcoin voluntary. | first-hand 2025-02-26 |
| IN-Q-TEL | entrenchment | 6 | operating — IQT, exempt since April 2001, reports 206 employees and net assets of $880,418,066 on its Form 990 for the year ending 31 March 2025, with revenue mainly from government grants. | official 2026-02-09 |
| IN-Q-TEL | funding insulation | 4 | graded funding disclosed — IQT's Form 990 for the tax year ending 31 March 2025 reports $114,227,920 of its $132,130,603 revenue as government grants (line 1e); the grantor agencies are not itemized. | official 2026-02-09 |
| IN-Q-TEL | reach | 4 | convenes or funds — Dreadnode funded-by In-Q-Tel (IQT) | official |
| IN-Q-TEL | reach | 4 | convenes or funds — IQT describes itself as set up in 1999 by the CIA as a not-for-profit strategic investment firm that takes small equity stakes and board-observer status in companies and lists its government partners, including the CIA, NSA and DHS. | first-hand |
| INTERNATIONAL NETWORK FOR ADVANCED AI MEASUREMENT, EVALUATION AND SCIENCE (NAAIMES) | entrenchment | 6 | operating — GOV.UK stated on 9 December 2025 that the Network became the International Network for Advanced AI Measurement, Evaluation and Science with the UK as Network Coordinator. | official 2025-12-09 |
| INTERNATIONAL NETWORK FOR ADVANCED AI MEASUREMENT, EVALUATION AND SCIENCE (NAAIMES) | reach | 4 | convenes or funds — The US Commerce fact sheet of 20 November 2024 launched the International Network of AI Safety Institutes with the US as inaugural chair and ten initial members as a forum bringing together technical expertise. | official 2024-11-20 |
| IRREGULAR | entrenchment | 4 | cited standard — Google DeepMind's Gemini 4 Argon report and Meta's Muse Spark report each use evaluation suites developed by Irregular (CyScenarioBench), and Irregular says the UK government and Anthropic use its SOLVE framework. | first-hand 2026-10-01 |
| IRREGULAR | funding insulation | 4 | graded funding disclosed — Irregular's 17 September 2025 release says it has reached millions in annual revenue and works side by side with labs like OpenAI and Anthropic; the only customers it names are the labs it evaluates and the UK government. | first-hand 2025-09-17 |
| IRREGULAR | reach | 6 | output acted on — Irregular (formerly Pattern Labs) evaluator-for Meta | official |
| IRREGULAR | reach | 6 | output acted on — Irregular (formerly Pattern Labs) evaluator-for OpenAI | official |
| IRREGULAR | reach | 6 | output acted on — Irregular (formerly Pattern Labs) evaluator-for DeepMind | official |
| IRREGULAR | reach | 6 | output acted on — Irregular (formerly Pattern Labs) evaluator-for Anthropic | official |
| IRREGULAR | reach | 6 | output acted on — Irregular (formerly Pattern Labs) contract UK AI Security Institute | official |
| IRREGULAR | reach | 5 | output acted on — Irregular (formerly Pattern Labs) evaluator-for Google / Alphabet | press |
| J. NATHAN MATIAS | durability | 6 | built operating — Matias founded the Citizens and Technology Lab (CAT Lab, from CivilServant), now at Cornell, which continues to publish. | first-hand 2017-01-01 |
| J. NATHAN MATIAS | insulation | 6 | independent money — CAT Lab states it 'does not accept funds from entities that we study' and lists NSF and foundation funders (MacArthur, Ford, Knight, Templeton, Heising-Simons). | first-hand |
| J. NATHAN MATIAS | leverage | 4 | advisory seat — Matias's CV lists Advisory Board, National Internet Observatory, Northeastern University (2021 to now) and Advisory Board, UN Human Development Report (2024-2025). | first-hand 2021-01-01 |
| JAAN TALLINN | durability | 6 | built operating — Tallinn co-founded the Future of Life Institute (2014), whose people page lists him as board member, and CSER (set up 2012); both continue to operate. | first-hand 2014-01-01 |
| JAAN TALLINN | durability | 6 | built operating — Jaan Tallinn co-founder Centre for the Study of Existential Risk (Cambridge) | official |
| JAAN TALLINN | insulation | 4 | appointed officer — Tallinn's own bios page says he 'serves on the Board of the Center for AI Safety'. | first-hand |
| JAAN TALLINN | leverage | 8 | signs grants — SFF states its Initiative Committee, 'made up of Jaan Tallinn, SFF Advisors, and 2-5 anonymous voters', makes grants on its own initiative without applications; SFF-2025 was run for Funder Jaan Tallinn. | first-hand |
| JAAN TALLINN | leverage | 4 | advisory seat — Jaan Tallinn board-seat Center for AI Safety | official |
| JAAN TALLINN | leverage | 4 | funds others — Epoch AI funded-by Jaan Tallinn | official |
| JAAN TALLINN | leverage | 4 | funds others — Faculty funded-by Jaan Tallinn | official |
| JAAN TALLINN | leverage | 4 | funds others — Machine Intelligence Research Institute (MIRI) funded-by Jaan Tallinn | official |
| JAAN TALLINN | leverage | 4 | funds others — LawZero funded-by Jaan Tallinn | official |
| JAAN TALLINN | leverage | 4 | funds others — Survival and Flourishing Fund funded-by Jaan Tallinn | official |
| JAAN TALLINN | leverage | 4 | funds others — Center for Humane Technology funded-by Jaan Tallinn | official |
| JAAN TALLINN | leverage | 4 | funds others — Seismic Catalyst Foundation funded-by Jaan Tallinn | official |
| JAAN TALLINN | leverage | 4 | funds others — Resolution funded-by Jaan Tallinn | official |
| JACK CLARK | durability | 6 | built operating — Clark is a co-founder of Anthropic and leads its Anthropic Institute, announced 11 March 2026. | first-hand 2026-03-11 |
| JACK CLARK | durability | 1 | publications — Jack Clark co-author Helen Toner | scholarly |
| JACK CLARK | durability | 1 | publications — Jack Clark co-author Miles Brundage | scholarly |
| JACK CLARK | durability | 1 | publications — Jack Clark co-author Logan Graham | scholarly |
| JACK CLARK | insulation | 4 | appointed officer — Anthropic's announcement says the Anthropic Institute 'is led by our co-founder Jack Clark, who will assume a new role as Anthropic's Head of Public Benefit'; no protection from removal stated. | first-hand 2026-03-11 |
| JACK CLARK | insulation | 3 | appointed officer — Jack Clark appointed-by US Department of Commerce | press |
| JACK CLARK | leverage | 4 | advisory seat — Jack Clark advisory Center for a New American Security | official |
| JACK CLARK | leverage | 4 | advisory seat — Jack Clark advisory Organisation for Economic Co-operation and Development (OECD) | official |
| JACK CLARK | leverage | 3 (ended) | advisory seat — The Commerce Department's first 27 National AI Advisory Committee members, as Nextgov printed them, include 'Jack Clark, Anthropic'. | press 2022-04-14 |
| JACK CLARK | leverage | 2 | public voice — Jack Clark testified U.S. House of Representatives | official |
| JACOB KLEIN | durability | 6 (ended) | built operating — Klein founded Coinbase's Trust and Safety organization and grew it into a multi-team function, per his CYBERUK 2026 bio. | first-hand |
| JACOB KLEIN | insulation | 2 | at will employee — Klein is an Anthropic employee heading a team; no term or protection stated. | first-hand |
| JACOB KLEIN | insulation | 2 | at will employee — Jacob Klein employed-by Coinbase | official |
| JACOB KLEIN | insulation | 2 | at will employee — Jacob Klein employed-by Google / Alphabet | official |
| JACOB KLEIN | leverage | 6 | runs org acted on — Klein is Head of Threat Intelligence at Anthropic, leading the team that detects and investigates AI misuse and feeds safeguards; Anthropic reports banning accounts and sharing indicators with authorities. | first-hand |
| JACOB STEINHARDT | durability | 6 | built operating — Steinhardt co-founded Transluce, which launched 23 October 2024 and continues to publish evaluations. | first-hand 2024-10-23 |
| JACOB STEINHARDT | durability | 3 | cited work — Steinhardt co-authored 'Measuring Massive Multitask Language Understanding' (ICLR 2021, arXiv 2009.03300) and 'Concrete Problems in AI Safety' (arXiv 1606.06565). | scholarly 2020-09-07 |
| JACOB STEINHARDT | insulation | 4 | appointed officer — Transluce's About page lists Steinhardt as Co-Founder, CEO and one of a three-person Board of Directors with Alex Allain and Mike McCormick. | first-hand |
| JACOB STEINHARDT | insulation | 2 | at will employee — Jacob Steinhardt employed-by OpenAI | official |
| JACOB STEINHARDT | insulation | 2 | at will employee — Jacob Steinhardt employed-by Open Philanthropy | official |
| JACOB STEINHARDT | leverage | 6 | runs org acted on — Steinhardt is Co-Founder and CEO of Transluce, which signed legal agreements giving it privileged access to OpenAI, Anthropic and Google DeepMind for its August 2026 mental health evaluation. | first-hand 2026-08-31 |
| JACOB STEINHARDT | leverage | 4 | advisory seat — Jacob Steinhardt advisory LawZero | official |
| JADE LEUNG | durability | 6 (ended) | built operating — GovAI's bio says she 'helped to co-found and lead GovAI as Head of Research & Partnerships'; GovAI continues to operate. | first-hand |
| JADE LEUNG | durability | 1 | publications — Jade Leung co-author Logan Graham | scholarly |
| JADE LEUNG | insulation | 4 | appointed officer — Leung holds the Vice-Chair seat by government appointment announced 23 September 2026; no removal protection stated. | official 2026-09-23 |
| JADE LEUNG | leverage | 6 (ended) | runs org acted on — As CTO of the AI Security Institute and the Prime Minister's AI adviser (reporting directly to the PM), she led the Institute's technical work until stepping back in September 2026. | official 2025-08-15 |
| JADE LEUNG | leverage | 4 | advisory seat — GOV.UK (23 Sep 2026): Leung becomes Vice-Chair of the AI Security Institute and Security Advisor to the AI Taskforce, providing strategic advice on AI safety and security. | official 2026-09-23 |
| JAIME SEVILLA | durability | 6 | built operating — Epoch's impact report says Epoch spun out in early 2025 as an independent 501(c)(3) with $10.3 million raised and 21 staff; Sevilla co-founded it in April 2022. | first-hand 2022-04-01 |
| JAIME SEVILLA | insulation | 4 | appointed officer — Form 990-EZ (FY2024) lists Jaime Sevilla Molina as Treasurer of Epoch Artificial Intelligence Inc at $0; he is also a director on a four-member board and CEO. | official 2024-12-31 |
| JAIME SEVILLA | leverage | 6 | runs org acted on — Sevilla is co-founder and CEO of Epoch AI, which its 2025 impact report says consulted UK AISI, EU AI Office, Sequoia and Bridgewater and briefed US congresspeople. | first-hand 2025-01-01 |
| JAMES MANYIKA | durability | 4 | cited work — The UN Advisory Body's Final Report 'Governing AI for Humanity' (Sept 2024), which he co-chaired, is published on the UN site. | official 2024-09-01 |
| JAMES MANYIKA | insulation | 2 | at will employee — Google's CEO placed Google Research under Manyika as a senior executive; no term or removal protection is stated. | first-hand 2023-04-20 |
| JAMES MANYIKA | leverage | 6 | runs org acted on — Google's CEO announced that Google Research 'will report in to James Manyika along with his existing Tech & Society teams'. | first-hand 2023-04-20 |
| JAMES MANYIKA | leverage | 4 (ended) | advisory seat — The UN biography lists James Manyika as Co-Chair of the Secretary-General's High-level Advisory Body on AI, whose final report 'Governing AI for Humanity' was released in September 2024. | official 2024-09-01 |
| JAN LEIKE | durability | 4 | cited work — Leike's own site says he developed OpenAI's approach to alignment research, co-authored the Superalignment Team's research roadmap, and at DeepMind prototyped reinforcement learning from human feedback. | first-hand |
| JAN LEIKE | insulation | 2 | at will employee — Leike's own site states 'Currently I'm working on a new project at Anthropic.' | first-hand |
| JAN LEIKE | leverage | 6 (ended) | runs org acted on — Leike's own site says 'Until the beginning of 2026, I led the Alignment team at Anthropic, where I was involved in scalable oversight research, character training, constitutional classifiers, and improving Claude's alignment', and that he previously co-led OpenAI's Superalignment Team. | first-hand |
| JAN LEIKE | leverage | 2 | public voice — Leike publishes the Aligned Substack; the January 2026 essay states alignment 'is an unsolved problem' while arguing it 'increasingly looks solvable' (cited by the page; not re-fetched). | first-hand 2026-01-01 |
| JARED KAPLAN | durability | 6 | built operating — Jared Kaplan co-founder Anthropic | official |
| JARED KAPLAN | durability | 3 | cited work — Kaplan is first author of 'Scaling Laws for Neural Language Models' (arXiv:2001.08361). | scholarly 2020-01-23 |
| JARED KAPLAN | durability | 2 | publications — Jared Kaplan co-author Joe Carlsmith | official |
| JARED KAPLAN | durability | 2 | publications — Jared Kaplan co-author Holden Karnofsky | official |
| JARED KAPLAN | durability | 1 | publications — Jared Kaplan co-author Dario Amodei | scholarly |
| JARED KAPLAN | durability | 1 | publications — Jared Kaplan co-author Amanda Askell | scholarly |
| JARED KAPLAN | durability | 1 | publications — Jared Kaplan co-author Chris Olah | scholarly |
| JARED KAPLAN | insulation | 4 | appointed officer — Anthropic: 'Co-Founder and Chief Science Officer Jared Kaplan will serve as Anthropic's Responsible Scaling Officer, succeeding Co-Founder and Chief Technology Officer Sam McCandlish.' | first-hand 2024-10-15 |
| JARED KAPLAN | insulation | 2 | at will employee — Jared Kaplan employed-by Johns Hopkins University | official |
| JARED KAPLAN | insulation | 1 | at will employee — Jared Kaplan employed-by OpenAI | press |
| JARED KAPLAN | leverage | 8 | signs deployment — Anthropic's Responsible Scaling Policy v3.0 gives the Responsible Scaling Officer duties including 'approving relevant model development or deployment decisions based on our risk assessments'; Kaplan holds the RSO seat. | first-hand 2026-02-24 |
| JASON MATHENY | durability | 6 | built operating — RAND's release says Matheny was founding director of the Center for Security and Emerging Technology at Georgetown University, which continues to operate. | official 2022-06-07 |
| JASON MATHENY | insulation | 4 | appointed officer — RAND's board chair announced the board selected Matheny as president and CEO after considering nearly 200 candidates. | official 2022-06-07 |
| JASON MATHENY | insulation | 1 | at will employee — Jason Matheny employed-by White House | press |
| JASON MATHENY | leverage | 6 | runs org acted on — RAND announced Jason Matheny president and CEO of the RAND Corporation, effective July 5, 2022 (the sixth president and CEO). | official 2022-06-07 |
| JASON MATHENY | leverage | 4 | advisory seat — Jason Matheny board-seat Anthropic | official |
| JASON MATHENY | leverage | 2 | public voice — Jason Matheny testified US Senate | official |
| JD VANCE | durability | 2 | publications — The White House biography calls him 'the author of the bestselling memoir Hillbilly Elegy'. | official |
| JD VANCE | insulation | 8 | statutory term — Constitution Art. II, Sec. 1 gives the President and Vice President a term of four years, and Sec. 4 allows removal only on impeachment for and conviction of treason, bribery, or other high crimes and misdemeanors. | official |
| JD VANCE | leverage | 10 | statutory authority — Constitution Art. I, Sec. 3: the Vice President 'shall be President of the Senate, but shall have no vote, unless they be equally divided', a binding casting vote. | official |
| JD VANCE | leverage | 4 | advisory seat — 50 U.S.C. 3021(c) lists the Vice President as a member of the National Security Council, whose function is to advise the President on integrating domestic, foreign and military policies relating to national security. | official |
| JEFFREY DING | durability | 1 | publications — Princeton University Press published his 'Technology and the Rise of Great Powers' on August 20, 2024, winner of the Lepgold Book Prize. | scholarly 2024-08-20 |
| JEFFREY DING | insulation | 2 | at will employee — Jeffrey Ding employed-by George Washington University | official |
| JEFFREY DING | leverage | 4 | advisory seat — GovAI's Annual Report 2025 lists Jeffrey Ding under its Board. | first-hand |
| JEFFREY DING | leverage | 4 | advisory seat — Jeffrey Ding advisory National Security Commission on Artificial Intelligence | official |
| JEFFREY DING | leverage | 4 | advisory seat — Jeffrey Ding board-seat Centre for the Governance of AI | official |
| JEFFREY DING | leverage | 2 | public voice — Jeffrey Ding testified U.S.-China Economic and Security Review Commission | official |
| JEFFREY DING | leverage | 1 | public voice — Jeffrey Ding testified US Senate | press |
| JEFFREY LADISH | durability | 6 | built operating — Palisade's About page says 'Jeffrey assembled a team to investigate emerging AI behavior and strategic capabilities' and lists him as Executive Director of the operating nonprofit. | first-hand |
| JEFFREY LADISH | insulation | 4 | appointed officer — Palisade Research Inc's Form 990 summary lists him as 'Pres/exec Dir' for fiscal 2024 ($117,578) and President for 2023. | official 2024-12-31 |
| JEFFREY LADISH | insulation | 2 | at will employee — Jeffrey Ladish employed-by Anthropic | official |
| JEFFREY LADISH | leverage | 2 | public voice — Palisade's About page says Ladish 'has helped advise the White House, Department of Defense, and congressional offices on risks from AI and emerging technologies' (no seat named). | first-hand |
| J.J. ALLAIRE | durability | 8 | program of record — UK AISI credits JJ Allaire as 'creator of Inspect'; the post says Inspect is used across AISI's frontier evals, by METR and Apollo Research, other governments, with 240+ contributors. | official 2026-06-18 |
| J.J. ALLAIRE | durability | 6 | built operating — J.J. Allaire founded Meridian Labs | official |
| J.J. ALLAIRE | durability | 6 | built operating — J.J. Allaire founded Posit PBC | official |
| J.J. ALLAIRE | insulation | 4 | appointed officer — Posit's page lists him 'Founder and Executive Chairman at Posit, PBC' (title only; no control stated). | first-hand |
| J.J. ALLAIRE | insulation | 2 | at will employee — J.J. Allaire employed-by Center for AI Standards and Innovation (NIST) | official |
| J.J. ALLAIRE | insulation | 2 | at will employee — J.J. Allaire employed-by UK AI Security Institute | official |
| J.J. ALLAIRE | leverage | 6 | runs org acted on — Meridian Labs lists him Principal Engineer and co-founder (2025); UK AISI says Inspect, which he created, 'powers many of AISI's frontier evaluations' and that METR and Apollo Research moved onto it. | official 2026-06-18 |
| JOANNE JANG | durability | 5 (ended) | built operating — The Model Behavior team she founded still operates inside OpenAI's Post Training group, and OpenAI's public Model Spec (18 Dec 2025 version) continues; the Spec page credits no author. | press 2025-12-18 |
| JOANNE JANG | insulation | 1 (ended) | at will employee — She left Model Behavior in Sep 2025 to found OAI Labs inside OpenAI and announced leaving OpenAI in Apr 2026; no term or protection stated in any source read. | press 2025-09-05 |
| JOANNE JANG | leverage | 8 (ended) | signs enforcement — Writing as 'I lead model behavior at OpenAI', she explained the Day 1 content policy for 4o image generation in ChatGPT (a shift from blanket refusals to harm-focused policy). | first-hand 2025-03-27 |
| JOANNE JANG | leverage | 5 (ended) | runs org acted on — TechCrunch: OpenAI's Model Behavior team (about 14 researchers) shapes how the company's models interact with people; its 'founding leader, Joanne Jang' moved to a new project in Sep 2025 and the team joined Post Training. | press 2025-09-05 |
| JOE CARLSMITH | durability | 4 | cited work — Anthropic's constitution page credits him: 'Joe Carlsmith wrote significant parts of many sections and played a core role in revising the text'; the document is released CC0. | official |
| JOE CARLSMITH | insulation | 2 | at will employee — He states he began a new role at Anthropic in mid-November 2025; no term or protection is stated. | first-hand 2025-11-03 |
| JOE CARLSMITH | leverage | 6 (ended) | runs org acted on — He states he joined Open Philanthropy's Worldview Investigations team in 2019 and 'eventually: led' it; he left in November 2025. | first-hand 2025-11-03 |
| JOE CARLSMITH | leverage | 2 | public voice — His own site announces he is joining Anthropic 'helping with the design of Claude's character/constitution/spec' and publishes his essays and talks. | first-hand 2025-11-03 |
| JOEL KAPLAN | insulation | 3 | appointed officer — Meta's Chief Global Affairs Officer; a Meta spokesperson told Reuters he 'will leave the company next year after 15 years there'. | press 2026-10-20 |
| JOEL KAPLAN | insulation | 2 | at will employee — Joel Kaplan employed-by OMB | official |
| JOEL KAPLAN | insulation | 2 | at will employee — Joel Kaplan employed-by Energy Future Holdings Corp. | official |
| JOEL KAPLAN | leverage | 8 | signs enforcement — Under the byline 'Joel Kaplan, Chief Global Affairs Officer', Meta announced it would end third-party fact-checking in the US and remove restrictions on topics like immigration and gender identity. | first-hand 2025-01-07 |
| JOHANN REHBERGER | durability | 4 | cited work — The same biography says he has contributed to the MITRE ATT&CK and ATLAS frameworks. | first-hand 2025-12-01 |
| JOHANN REHBERGER | insulation | 2 | at will employee — Johann Rehberger employed-by Electronic Arts | official |
| JOHANN REHBERGER | insulation | 2 | at will employee — Johann Rehberger employed-by Microsoft | official |
| JOHANN REHBERGER | leverage | 2 | public voice — His Embrace The Red write-up reports remote code execution in GitHub Copilot, reported to Microsoft 29 June 2025, assigned CVE-2025-53773 and fixed in the August 2025 Patch Tuesday release. | first-hand 2025-06-29 |
| JOSH JACOBSON | insulation | 4 | appointed officer — MATS Research Inc's Form 990 summary for fiscal 2024 lists Joshua Jacobson as President ($0). | official 2024-12-31 |
| JOSH JACOBSON | insulation | 2 | at will employee — Josh Jacobson employed-by FAR.AI | official |
| JOSH JACOBSON | insulation | 2 | at will employee — Josh Jacobson employed-by Berkeley Existential Risk Initiative (BERI) | official |
| JOSH JACOBSON | insulation | 2 | at will employee — Josh Jacobson employed-by CEA | official |
| JOSH JACOBSON | insulation | 2 | at will employee — Josh Jacobson employed-by METR (Model Evaluation & Threat Research) | official |
| JOSH JACOBSON | leverage | 4 | advisory seat — FAR.AI's author page titles him 'Board Advisor' and says he 'remained involved in an advisory capacity' after serving as interim COO in 2022 and early 2023. | first-hand |
| JOSH JACOBSON | leverage | 4 | advisory seat — Josh Jacobson board-seat MATS Research | official |
| JUNIPER DOWNS | insulation | 2 | at will employee — Airbnb hired her as Global Head of Community Policy and Partnerships; no term or protection is stated. | first-hand 2021-10-07 |
| JUNIPER DOWNS | leverage | 8 | signs enforcement — Airbnb announced she 'will oversee the team responsible for the rules that govern behavior on the Airbnb platform' as Global Head of Community Policy and Partnerships; in March 2024 she is quoted announcing the new security-camera rules. | first-hand 2021-10-07 |
| JUNIPER DOWNS | leverage | 2 | public voice — Juniper Downs testified U.S. House of Representatives | official |
| JUNIPER DOWNS | leverage | 2 | public voice — Juniper Downs testified US Senate | official |
| KAI FRONSDAL | durability | 6 | built operating — Meridian's post says he developed Petri and that it now continues at Meridian; the UK AI Security Institute built its alignment evaluation pipeline on it and Anthropic uses it in its own assessments. | first-hand 2026-05-07 |
| KAI FRONSDAL | insulation | 2 | at will employee — Kai Fronsdal employed-by Meridian Labs | official |
| KAI FRONSDAL | leverage | 4 | advisory seat — Kai Fronsdal advisory UK AI Security Institute | official |
| KAI FRONSDAL | leverage | 2 | public voice — First author of Anthropic's Petri alignment-auditing post (6 Oct 2025), written as an Anthropic Fellow. | first-hand 2025-10-06 |
| KATE KLONICK | durability | 2 | publications — Kate Klonick co-author Renee DiResta | official |
| KATE KLONICK | durability | 2 | publications — Kate Klonick co-author Yoel Roth | official |
| KATE KLONICK | durability | 1 | publications — Author of 'The New Governors', Harvard Law Review vol. 131 (2018), and 'The Facebook Oversight Board', 129 Yale L.J. 2418 (2020). | scholarly 2020-02-01 |
| KATE KLONICK | insulation | 6 | tenure — Her CV lists her as Associate Professor with Tenure at St. John's University School of Law, 2022 to present. | first-hand 2025-08-19 |
| KATE KLONICK | insulation | 2 | at will employee — Kate Klonick employed-by Lawfare | official |
| KATE KLONICK | leverage | 4 | advisory seat — Kate Klonick advisory Electronic Privacy Information Center | official |
| KATE KLONICK | leverage | 2 | public voice — Kate Klonick testified UK House of Lords Communications and Digital Select Committee | official |
| KATE KLONICK | leverage | 2 | public voice — Kate Klonick testified U.S. House of Representatives | official |
| KATE STARBIRD | durability | 6 | built operating — The UW Center for an Informed Public's people page lists Starbird as 'CIP Co-founder | Professor, UW Department of Human Centered Design & Engineering'. | first-hand |
| KATE STARBIRD | insulation | 0 (ended) | removed not reinstated — Acting DHS Secretary Huffman's 20 January 2025 memo states: 'I am directing the termination of all current memberships on the advisory committees within DHS, effective immediately' (committee-wide, not aimed at her; the CISA subcommittee she chaired had already disbanded). | press 2025-01-20 |
| KATE STARBIRD | leverage | 4 (ended) | advisory seat — CISA's readout of its 22 June 2022 Cybersecurity Advisory Committee meeting lists the MDM subcommittee as presented by Dr. Kate Starbird, whose chair recommended that CISA 'invest in external research to assess the impact of MDM threats'. | official 2022-06-22 |
| KATE STARBIRD | leverage | 2 | public voice — UW HCDE's faculty page describes Starbird as a Professor of Human Centered Design & Engineering whose published research follows online rumoring and disinformation. | official |
| KELLIE LIKET | durability | 6 | built operating — Kellie Liket co-founder Ergo Impact Inc | official |
| KELLIE LIKET | insulation | 4 | appointed officer — Ergo Impact's TY2025 Form 990 lists her as President (40 hours a week, $325,653 reportable compensation). | official 2026-01-01 |
| KELLIE LIKET | leverage | 4 | advisory seat — Kellie Liket board-seat Sentinel Bio | official |
| KELLIE LIKET | leverage | 4 | advisory seat — Kellie Liket board-seat Seismic Catalyst Foundation | official |
| KELLIE LIKET | leverage | 4 | advisory seat — Kellie Liket board-seat Halcyon Futures | official |
| KELLIE LIKET | leverage | 4 | advisory seat — Kellie Liket board-seat Fathom AI Inc | official |
| KELLIE LIKET | leverage | 4 | funds others — Ergo Impact, of which she is President, reported $10,000 in grants paid on its TY2025 Form 990 (program: finding, funding and scaling promising people and solutions). | official 2026-01-01 |
| KENT WALKER | durability | 2 | publications — Posts under his byline on Google's blog, such as the 21 Nov 2024 DOJ search-remedies post, remain available. | first-hand 2024-11-21 |
| KENT WALKER | insulation | 4 | appointed officer — Alphabet's 2026 proxy lists him as an executive officer, President Global Affairs and CLO since November 2021 and Secretary since January 2020. | official 2026-04-24 |
| KENT WALKER | leverage | 8 | signs enforcement — Alphabet's 2026 proxy says he 'oversees teams responsible for content policy and responsible AI development' as President, Global Affairs and Chief Legal Officer. | official 2026-04-24 |
| KEVIN ESVELT | durability | 6 | built operating — His own Substack says he co-founded SecureBio and SecureDNA; SecureBio's site describes SecureDNA as developing screening software for DNA synthesis orders for providers and device makers. | first-hand 2026-04-29 |
| KEVIN ESVELT | insulation | 6 | tenure — MIT Media Lab announced Esvelt was granted tenure, effective 1 July 2026. | first-hand 2026-06-15 |
| KEVIN ESVELT | leverage | 4 | advisory seat — Kevin Esvelt board-seat SecureBio | official |
| KEVIN ESVELT | leverage | 2 | public voice — Kevin Esvelt testified US Senate | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | entrenchment | 6 | operating — The foundation filed its Form 990-PF for tax year 2024 on 6 November 2025 reporting $179,620,479 in charitable disbursements. | official 2025-11-06 |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | funding insulation | 8 | endowment or appropriation — Knight's Form 990-PF for 2024 reports total assets of $2,533,415,674, revenue of $139,912,312, contributions received of $0 and charitable disbursements of $179,620,479. | official 2025-11-06 |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Knight-Georgetown Institute (KGI) funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Berkman Klein Center for Internet & Society (Harvard) funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Global Disinformation Index (Disinformation Index Ltd, UK; Disinformation Index, Inc., US) funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Heritage Foundation funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Carnegie Mellon University funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — NYU Center for Social Media and Politics funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — University of Washington Center for an Informed Public funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Knight First Amendment Institute at Columbia University funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Partnership on AI funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Press Forward funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Election Integrity Partnership funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Aspen Institute funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — NewsGuard funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — International Network for Advanced AI Measurement, Evaluation and Science (formerly International Network of AI Safety Institutes) funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — SeedAI funded-by John S. and James L. Knight Foundation | official |
| JOHN S. AND JAMES L. KNIGHT FOUNDATION | reach | 4 | convenes or funds — Kate Klonick funded-by John S. and James L. Knight Foundation | official |
| KYLE SCOTT | insulation | 4 | appointed officer — ARC's team page: 'ARC's officers are Paul Christiano (President), Kyle Scott (Treasurer), and Harshita Khera (Secretary)'. | first-hand |
| KYLE SCOTT | insulation | 2 | at will employee — Kyle Scott employed-by METR (Model Evaluation & Threat Research) | official |
| KYLE SCOTT | leverage | 4 | advisory seat — Kyle Scott board-seat Palisade Research | official |
| KYLE SCOTT | leverage | 4 | advisory seat — Kyle Scott board-seat Alignment Research Center | official |
| LARRY FINK | durability | 6 | built operating — The same proxy says he has served as Chairman since founding BlackRock in 1988; BlackRock reported $15.3 trillion in AUM for Q2 2026. | official 2026-04-10 |
| LARRY FINK | insulation | 4 | appointed officer — BlackRock's 2026 proxy: 'The Board has re-appointed Laurence D. Fink as its Chairman'; the filing text read states no voting control. | official 2026-04-10 |
| LARRY FINK | leverage | 4 | funds others — As Chairman and CEO of BlackRock he is quoted announcing the Global AI Infrastructure Investment Partnership with GIP, MGX and Microsoft to invest in data centers and power. | official 2024-09-17 |
| LATCHBIO | entrenchment | 6 | operating — LatchBio, Inc. operates from San Francisco with roughly 100 staff and 5,000 platform scientists, per its own job postings. | first-hand 2026-08-29 |
| LATCHBIO | entrenchment | 4 | cited standard — LatchBio's postings state frontier labs use its benchmarks and that its work appears in Anthropic and xAI model cards. | first-hand |
| LATCHBIO | funding insulation | 6 | independent funding — LatchBio's own postings list its backers as Lux Capital, Coatue, Fifty Years, Haystack, Jeff Dean, Alec Nielsen and George Church; none is a lab it grades. | first-hand |
| LATCHBIO | reach | 6 | output acted on — LatchBio evaluator-for Anthropic | official |
| LATCHBIO | reach | 6 | output acted on — LatchBio evaluator-for xAI | official |
| LATCHBIO | reach | 6 | output acted on — xAI's Grok 4.7 post says the model tops "LatchBio's biosafety benchmark at 62.4%". | first-hand 2026-09-21 |
| LAWRENCE CHAN | durability | 4 | cited work — CCST says he co-led METR's 'time horizon' research on AI agents, covered in Nature and 'a widely used metric for tracking AI progress'. | official 2026-08-19 |
| LAWRENCE CHAN | insulation | 4 | appointed officer — FAR.AI's 2024 Form 990 lists Lawrence Chan as Secretary; FAR.AI's team page lists him as Board Member. | official 2024-12-31 |
| LAWRENCE CHAN | insulation | 2 | at will employee — Lawrence Chan employed-by California Governor's Office of Emergency Services (Cal OES) | official |
| LAWRENCE CHAN | insulation | 2 | at will employee — Lawrence Chan employed-by METR (Model Evaluation & Threat Research) | official |
| LAWRENCE CHAN | insulation | 2 | at will employee — Lawrence Chan employed-by Redwood Research | official |
| LAWRENCE CHAN | leverage | 4 | advisory seat — California Council on Science and Technology: he is Science Advisor to Cal OES and 'will advise senior leadership on frontier AI safety and risk assessment'. | official 2026-08-19 |
| LAWRENCE CHAN | leverage | 4 | advisory seat — Lawrence Chan board-seat FAR.AI | official |
| LAWZERO | entrenchment | 6 | operating — LawZero operates as a Montreal nonprofit with a published team, board and donor list. | first-hand |
| LAWZERO | funding insulation | 8 | endowment or appropriation — Canada and Germany announced funding of up to CAD 300 million for LawZero; Canada plans CAD 150 million via the Strategic Response Fund and Germany EUR 100 million, subject to European Commission notification. | official 2026-09-16 |
| LAWZERO | reach | 2 | publishes — LawZero publishes research on its Scientist AI safe-by-design approach. | first-hand |
| LEONARD TANG | durability | 5 | built operating — Leonard Tang founded Haize Labs | press |
| LEONARD TANG | insulation | 1 | at will employee — Leonard Tang employed-by Beacon Software | press |
| LEONARD TANG | leverage | 6 (ended) | runs org acted on — As co-founder and CEO of Haize Labs, he led an organisation named among the red-teaming organizations in OpenAI's o1 system card. | first-hand 2024-12-05 |
| LEONARD TANG | leverage | 1 | public voice — Interviewed on Haize Labs' automated jailbreaking work (VentureBeat, June 2024). | press 2024-06-01 |
| LMARENA (ARENA) | entrenchment | 6 | operating — Arena Intelligence, Inc. operates arena.ai and reports over $100M annualized revenue. | first-hand 2026-10-08 |
| LMARENA (ARENA) | entrenchment | 4 | cited standard — The Arena leaderboard is the ranking labs quote in launch posts (Google for Gemini 3, xAI for Grok 4.1). | first-hand |
| LMARENA (ARENA) | funding insulation | 6 | independent funding — Arena announced a $200 million Series B at a $3.1 billion valuation co-led by Lightspeed and Khosla; the post names no customer and no revenue split by customer type. | first-hand 2026-10-08 |
| LMARENA (ARENA) | reach | 6 | output acted on — Google's Gemini 3 launch post says the model "tops the LMArena Leaderboard with a breakthrough score of 1501 Elo". | first-hand |
| LOGAN GRAHAM | durability | 6 | built operating — Anthropic's Frontier Red Team, which he built, still operates and publishes (its page lists posts through October 2026). | first-hand 2026-10-12 |
| LOGAN GRAHAM | insulation | 2 | at will employee — Logan Graham employed-by UK Government | official |
| LOGAN GRAHAM | insulation | 2 | at will employee — Logan Graham employed-by Anthropic | official |
| LOGAN GRAHAM | insulation | 2 | at will employee — Anthropic employee (Head of Frontier Red Team); no term or protection stated; the Cabinet Office record shows he joined Anthropic in Nov 2022. | official 2022-11-15 |
| LOGAN GRAHAM | leverage | 6 | runs org acted on — Head of Anthropic's Frontier Red Team; his CV says he 'built and directed a team of 15 researchers', and his testimony form states the team's pre-deployment testing partnerships with US CAISI and UK AISI. | first-hand 2025-12-17 |
| LOGAN GRAHAM | leverage | 2 | public voice — Logan Graham testified U.S. House of Representatives | official |
| LONGVIEW PHILANTHROPY | entrenchment | 6 | operating — Longview Philanthropy USA Inc is a 501(c)(3) public charity with exemption issued February 2024 and returns for 2024 and 2025. | official |
| LONGVIEW PHILANTHROPY | funding insulation | 2 | graded funding undisclosed — Longview's TY2025 Schedule B prints its contributor as RESTRICTED with no name; contributions were $20,927,481 of $21,167,086 revenue, and it advises the donors it serves. | official 2026-09-18 |
| LONGVIEW PHILANTHROPY | reach | 4 | convenes or funds — Jeffrey Ding funded-by Longview Philanthropy | official |
| LONGVIEW PHILANTHROPY | reach | 4 | convenes or funds — Longview's TY2025 Form 990 reports grants and similar amounts paid of $12,599,230 from its internal funds. | official 2026-09-18 |
| LONGVIEW PHILANTHROPY | reach | 4 | convenes or funds — Horizon Institute for Public Service funded-by Longview Philanthropy | official |
| LONGVIEW PHILANTHROPY | reach | 4 | convenes or funds — MATS Research funded-by Longview Philanthropy | official |
| LONGVIEW PHILANTHROPY | reach | 4 | convenes or funds — METR (Model Evaluation & Threat Research) funded-by Longview Philanthropy | official |
| LONGVIEW PHILANTHROPY | reach | 4 | convenes or funds — RAND funded-by Longview Philanthropy | official |
| LUKAS PETERSSON | durability | 6 | built operating — Lukas Petersson founded Andon Labs | official |
| LUKAS PETERSSON | durability | 6 | built operating — Co-founder of Andon Labs and co-author of Vending-Bench, which Anthropic's system card reports results on; the company continues to operate. | first-hand |
| LUKAS PETERSSON | durability | 4 | cited work — Vending-Bench (Backlund and Petersson, arXiv:2502.15840) is cited in Anthropic's Opus 4.6 system card. | first-hand 2026-02-05 |
| LUKAS PETERSSON | leverage | 6 | runs org acted on — Founder and CEO of Andon Labs (his CV); Anthropic's Claude Opus 4.6 system card has a section 'External testing from Andon Labs', and Andon says Vending-Bench is used at every major model release. | first-hand 2026-02-05 |
| LUKE MUEHLHAUSER | durability | 2 | publications — Author of 'Our AI Governance Grantmaking So Far' (8 Dec 2020) setting the funder's AI-governance priorities. | first-hand 2020-12-08 |
| LUKE MUEHLHAUSER | insulation | 2 | at will employee — Luke Muehlhauser employed-by Open Philanthropy | official |
| LUKE MUEHLHAUSER | insulation | 2 | at will employee — Luke Muehlhauser employed-by Machine Intelligence Research Institute (MIRI) | official |
| LUKE MUEHLHAUSER | leverage | 4 | advisory seat — Luke Muehlhauser board-seat Anthropic | official |
| LUKE MUEHLHAUSER | leverage | 4 | funds others — Coefficient Giving's team page: Managing Director, AI Governance & International Policy; 'He leads our grantmaking on AI governance and international policy.' | first-hand |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | entrenchment | 6 | operating — MacArthur has operated since 1970 and reports grants of more than $9.19 billion since 1978. | first-hand |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | funding insulation | 8 | endowment or appropriation — MacArthur's 2024 Form 990-PF shows net assets of $8,650,796,551 and contributions received of $0. | official 2025-11-12 |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Partnership on AI funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Aspen Institute funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Berkman Klein Center for Internet & Society (Harvard) funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Brookings funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Institute for Strategic Dialogue funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Knight First Amendment Institute at Columbia University funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Press Forward funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Humanity AI funded-by John D. and Catherine T. MacArthur Foundation | official |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — MacArthur's Technology in the Public Interest program lists 247 grants and $70.5M to 114 organizations since 2017, including the Partnership on AI and AI Now Institute. | first-hand |
| JOHN D. AND CATHERINE T. MACARTHUR FOUNDATION | reach | 4 | convenes or funds — Kate Klonick funded-by John D. and Catherine T. MacArthur Foundation | official |
| MALA KUMAR | durability | 4 (ended) | cited work — As Director of Program Management, AI Safety at MLCommons she oversaw the core team that created its inaugural AI safety benchmark (AILuminate). | first-hand 2024-12-31 |
| MALA KUMAR | durability | 2 | publications — Mala Kumar co-author Rumman Chowdhury | official |
| MALA KUMAR | durability | 1 | publications — Mala Kumar co-author Peter Mattson | scholarly |
| MALA KUMAR | durability | 1 | publications — Co-author of the OB-CAIE evaluation methodology paper, arXiv:2610.00529, September 30 2026. | scholarly 2026-09-30 |
| MALA KUMAR | insulation | 4 | appointed officer — Humane Intelligence's team page states she was named Executive Director in April 2026 after serving as Interim Executive Director from August 2025. | first-hand 2026-04-01 |
| MALA KUMAR | insulation | 2 | at will employee — Mala Kumar employed-by Humane Intelligence | official |
| MALA KUMAR | insulation | 2 | at will employee — Mala Kumar employed-by MLCommons | official |
| MALA KUMAR | leverage | 6 | runs org acted on — Mala Kumar is Executive Director of Humane Intelligence, which sells red teaming events and contextual evaluations as paid services to organizations. | first-hand 2026-04-01 |
| MALA KUMAR | leverage | 4 | convenes — Humane Intelligence, which she leads, runs red teaming events and worked with more than 200 organizations in 2025. | first-hand 2025-12-31 |
| MARC ANDREESSEN | durability | 6 | built operating — Andreessen Horowitz, which he co-founded in July 2009 (per Coinbase's proxy), operates today and closed a $15B raise in January 2026. | first-hand 2026-01-09 |
| MARC ANDREESSEN | durability | 4 | cited work — The a16z Techno-Optimist Manifesto, published under his name, remains on a16z.com and states 'We believe any deceleration of AI will cost lives.' | first-hand 2023-10-16 |
| MARC ANDREESSEN | insulation | 4 | appointed officer — Meta's 2026 proxy lists Andreessen as an independent director since 2008 (Risk & Strategy committee); Meta is controlled by its founder's dual-class stock, so the seat is removable by the controlling holder. | first-hand 2026-04-16 |
| MARC ANDREESSEN | leverage | 4 | advisory seat — The White House named Marc Andreessen among 13 appointees to the President's Council of Advisors on Science and Technology, co-chaired by David Sacks and Michael Kratsios. | official 2026-03-25 |
| MARC ANDREESSEN | leverage | 4 | funds others — a16z announced it raised over $15B, including 'over 18% of all venture capital dollars allocated in the United States in 2025', aimed at 'winning the key architectures of the future - AI and crypto'. | first-hand 2026-01-09 |
| MARC ANDREESSEN | leverage | 4 | funds others — Leading the Future (super PAC, FEC C00916114) funded-by Marc Andreessen | official |
| MARC ANDREESSEN | leverage | 4 | funds others — Fairshake (super PAC, FEC C00835959) funded-by Marc Andreessen | official |
| MARC ANDREESSEN | leverage | 4 | funds others — Right for America (super PAC, FEC C00867036) funded-by Marc Andreessen | official |
| MARC ANDREESSEN | leverage | 4 | funds others — SMP (Senate Majority PAC, FEC C00484642) funded-by Marc Andreessen | official |
| MARC ANDREESSEN | leverage | 4 | funds others — MAGA Inc. (super PAC, FEC C00892471) funded-by Marc Andreessen | official |
| MARCO MASCORRO | insulation | 2 | at will employee — Marco Mascorro employed-by Andreessen Horowitz (a16z) | official |
| MARCO MASCORRO | leverage | 4 | advisory seat — Marco Mascorro advisory METR (Model Evaluation & Threat Research) | official |
| MARCO MASCORRO | leverage | 4 (ended) | funds others — As an a16z partner he co-authored the announcement leading the Series A in Deeptune, an RL-environments company. | first-hand 2026-03-19 |
| MARCO MASCORRO | leverage | 4 | funds others — Mechanize funded-by Marco Mascorro | official |
| MARIUS HOBBHAHN | durability | 6 | built operating — Apollo Research, which he co-founded and leads as CEO, operates as a US public benefit corporation and runs evaluations and a monitoring product. | first-hand 2026-01-20 |
| MARIUS HOBBHAHN | insulation | 4 | appointed officer — Companies House lists him as an active director of Apollo Research AI Ltd (15289159), appointed 16 November 2023. | official 2023-11-16 |
| MARIUS HOBBHAHN | insulation | 2 | at will employee — Marius Hobbhahn employed-by Epoch AI | official |
| MARIUS HOBBHAHN | leverage | 6 | runs org acted on — OpenAI's o1 system card reports that Apollo Research, an evaluation organization, evaluated o1 for scheming, and reproduces Apollo's findings and recommendation. | official 2024-12-05 |
| MARIUS HOBBHAHN | leverage | 4 | advisory seat — Marius Hobbhahn advisory European AI Office | official |
| MARIUS HOBBHAHN | leverage | 2 | public voice — Marius Hobbhahn testified US Senate | official |
| MARK CARNEY | leverage | 8 | holds policy pen — The Prime Minister's guide Open and Accountable Government states that Ministers of the Crown 'are chosen by the Prime Minister' and 'serve at the pleasure of the Prime Minister', that 'Government policy is established by the Cabinet', and that all Governor-in-Council appointment recommendations are subject to the Prime Minister's approval. | official 2015-11-27 |
| MARK CARNEY | leverage | 3 | advisory seat — Mark Carney board-seat Bloomberg L.P. | press |
| MARK CARNEY | leverage | 2 (ended) | public voice — Delivered 'Breaking the tragedy of the horizon' at Lloyd's of London proposing an industry-led Climate Disclosure Task Force to deliver a voluntary disclosure standard. | first-hand 2015-09-29 |
| MARK CARNEY | leverage | 2 | public voice — Mark Carney is Canada's 24th Prime Minister, sworn in March 2025. | official 2025-03-01 |
| MARK ZUCKERBERG | durability | 8 | program of record — Meta's Q2 2026 results release (Exhibit 99.1 to Form 8-K) reports family daily active people of 3.60 billion on average for June 2026. | official 2026-07-29 |
| MARK ZUCKERBERG | insulation | 10 | voting control — Meta's 10-K states that Mark Zuckerberg, founder, Chairman and CEO, 'is able to exercise voting rights with respect to a majority of the voting power of our outstanding capital stock and therefore has the ability to control the outcome of all matters submitted to our stockholders for approval, including the election of directors and any merger, consolidation, or sale of all or substantially all of our assets.' | official 2026-01-29 |
| MARK ZUCKERBERG | leverage | 10 | controls deployer — Meta's 10-K states that Mark Zuckerberg, founder, Chairman and CEO, 'is able to exercise voting rights with respect to a majority of the voting power of our outstanding capital stock and therefore has the ability to control the outcome of all matters submitted to our stockholders for approval, including the election of directors and any merger, consolidation, or sale of all or substantially all of our assets.' Meta ships its Llama and Muse models to the public. | official 2026-01-29 |
| MARK ZUCKERBERG | leverage | 6 | runs org acted on — In his 23 July 2024 open-source letter Zuckerberg says 'Amazon, Databricks, and NVIDIA are launching full suites of services to support developers fine-tuning and distilling their own models' on Llama, the model family Meta releases openly. | first-hand 2024-07-23 |
| MARKUS ANDERLJUNG | durability | 2 | publications — Markus Anderljung co-author Conrad Stosz | official |
| MARKUS ANDERLJUNG | durability | 2 | publications — GovAI team page lists him as Principal Researcher leading research at GovAI, with papers in Science and Nature Machine Intelligence. | first-hand |
| MARKUS ANDERLJUNG | durability | 1 | publications — Markus Anderljung co-author Logan Graham | scholarly |
| MARKUS ANDERLJUNG | durability | 1 | publications — Markus Anderljung co-author Jade Leung | scholarly |
| MARKUS ANDERLJUNG | durability | 1 | publications — Markus Anderljung co-author Miles Brundage | scholarly |
| MARKUS ANDERLJUNG | insulation | 4 | appointed officer — Markus Anderljung appointed-by European AI Office | official |
| MARKUS ANDERLJUNG | insulation | 2 | at will employee — Markus Anderljung employed-by Centre for the Governance of AI | official |
| MARKUS ANDERLJUNG | insulation | 2 | at will employee — Markus Anderljung employed-by UK Government | official |
| MARKUS ANDERLJUNG | leverage | 4 | advisory seat — Markus Anderljung advisory European Commission | official |
| MARKUS ANDERLJUNG | leverage | 4 (ended) | advisory seat — The European Commission's Meet the Chairs page lists Markus Anderljung as Vice Chair of Working Group 4 of the General-Purpose AI Code of Practice. | official 2025-07-01 |
| MARKUS ANDERLJUNG | leverage | 4 | advisory seat — Markus Anderljung advisory Center for a New American Security | official |
| MARKUS ANDERLJUNG | leverage | 4 | advisory seat — Markus Anderljung advisory Organisation for Economic Co-operation and Development (OECD) | official |
| MARKUS ANDERLJUNG | leverage | 1 | public voice — Lead author of Frontier AI Regulation: Managing Emerging Risks to Public Safety (arXiv:2307.03718, July 2023). | scholarly 2023-07-06 |
| MATS RESEARCH | entrenchment | 6 | operating — MATS Research is an independent 501(c)(3) public charity (EIN 99-0648563) operating the program. | first-hand |
| MATS RESEARCH | funding insulation | 6 | independent funding — MATS's transparency page lists 2025 gifts from Coefficient Giving ($11,823,980; $3,629,280) and Good Ventures ($1,520,150; $3,038,300), none from a lab or evaluator it places fellows with (one funder exceeds 20 percent). | first-hand 2025-07-23 |
| MATS RESEARCH | reach | 4 | convenes or funds — MATS runs a ten to twelve week fellowship in Berkeley and London with funded extensions and reports 631 participants since late 2021. | first-hand |
| MATT FREDRIKSON | durability | 6 | built operating — Gray Swan's About page lists Matt Fredrikson as Co-founder, Chief Executive Officer; the company announced a Series A on 28 May 2026. | first-hand 2026-05-28 |
| MATT FREDRIKSON | durability | 3 | cited work — Co-author of Universal and Transferable Adversarial Attacks on Aligned Language Models (arXiv:2307.15043). | scholarly 2023-07-27 |
| MATT FREDRIKSON | durability | 1 | publications — Matt Fredrikson co-author Zico Kolter | scholarly |
| MATT FREDRIKSON | insulation | 2 | at will employee — Matt Fredrikson employed-by Carnegie Mellon University | official |
| MATT FREDRIKSON | leverage | 6 | runs org acted on — OpenAI's o1 system card lists Gray Swan AI among its red teaming organizations; Gray Swan, which he leads as CEO, is named as evaluator in frontier model documents. | official 2024-12-05 |
| MATT TAIBBI | durability | 2 | publications — Twitter Files threads, including Hamilton 68 (January 27 2023) and #19 (March 17 2023), remain posted. | first-hand 2023-03-17 |
| MATT TAIBBI | insulation | 6 | independent money — Racket News states it is 100% subscriber-supported with no alternative sources of financing. | first-hand |
| MATT TAIBBI | leverage | 2 | public voice — Gave written testimony to the House Weaponization Subcommittee on March 9 2023, stating 'For every government agency scanning Twitter, there were perhaps 20 quasi-private entities doing the same'. | official 2023-03-09 |
| MAX HENDERSON | durability | 6 | built operating — Ergo Impact Inc (EIN 99-2339731, exemption July 2024) filed a TY2025 Form 990 with $3.73 million of contributions. | official 2026-01-01 |
| MAX HENDERSON | durability | 6 | built operating — Max Henderson co-founder Ergo Impact Inc | official |
| MAX HENDERSON | insulation | 4 | appointed officer — Ergo Impact's TY2025 Form 990 names him principal officer and Chief Executive Officer, reporting $676,000 compensation. | official 2026-01-01 |
| MAX HENDERSON | insulation | 2 | at will employee — Max Henderson employed-by Halcyon Futures | official |
| MAX HENDERSON | leverage | 4 | advisory seat — Max Henderson board-seat Sentinel Bio | official |
| MAX HENDERSON | leverage | 4 | advisory seat — Max Henderson board-seat Fathom AI Inc | official |
| MAX HENDERSON | leverage | 4 | advisory seat — Max Henderson board-seat AI Verification and Evaluation Research Institute (AVERI) | official |
| MAX HENDERSON | leverage | 4 | funds others — Sentinel Bio funded-by Max Henderson | official |
| MAX HENDERSON | leverage | 4 | funds others — Ergo Impact, which he co-founded and leads as CEO, describes itself as finding, funding and scaling people and solutions. | first-hand |
| MAXIME LABONNE | durability | 4 | cited work — His llm-course GitHub repository stands at 84.7k stars on 24 October 2026. | first-hand 2026-10-24 |
| MAXIME LABONNE | insulation | 2 | at will employee — Maxime Labonne employed-by Liquid AI | official |
| MAXIME LABONNE | leverage | 2 | public voice — Published 'Uncensor any LLM with abliteration' on the Hugging Face blog, June 13 2024. | first-hand 2024-06-13 |
| MECHANIZE | entrenchment | 6 | operating — Mechanize describes itself as a team of around 35 people building environments and evals for frontier coding agents. | first-hand |
| MECHANIZE | funding insulation | 6 | independent funding — Mechanize raised $9.1 million at a $500 million post-money valuation from three named individual investors; no customer or revenue is stated. | first-hand 2026-04-24 |
| MECHANIZE | reach | 2 | publishes — Mechanize publishes GBA Eval, a benchmark of coding agents writing a Game Boy Advance emulator, and a blog of essays. | first-hand |
| MERIDIAN LABS | entrenchment | 6 | operating — Meridian Research Labs Inc is a 501(c)(3) with exemption issued March 2025 maintaining Inspect and Petri. | official 2025-03-01 |
| MERIDIAN LABS | entrenchment | 4 | cited standard — Inspect AI is used by UK AISI, US CAISI, METR, Apollo, Epoch, SecureBio, Redwood and RAND per Meridian's own pages; Anthropic donated Petri to be maintained by Meridian. | first-hand 2026-05-07 |
| MERIDIAN LABS | reach | 6 | output acted on — Meridian's About page says Inspect AI is the standard framework for UK AISI, US CAISI, EU AI Office, Japan AISI and Korea AISI, and Anthropic handed Petri development to Meridian Labs, whose runs appear in Anthropic model cards. | first-hand |
| META | entrenchment | 8 | program of record — Meta's Q2 2026 results report 3.60 billion daily active people on average for June 2026 and $60.80 billion quarterly revenue. | first-hand 2026-07-29 |
| META | funding insulation | 10 | controlled or statutory — Meta's 10-K for 2025 names control of a majority of voting power of its capital stock by its founder, Chairman and CEO, through a dual class structure. | first-hand 2026-01-29 |
| META | reach | 4 | convenes or funds — Carnegie Endowment for International Peace funded-by Meta | official |
| META | reach | 4 | convenes or funds — Gillian Hadfield funded-by Meta | official |
| META | reach | 4 | convenes or funds — Partnership on AI funded-by Meta | official |
| META | reach | 4 | convenes or funds — Scale AI funded-by Meta | official |
| META | reach | 4 | convenes or funds — SeedAI funded-by Meta | official |
| META | reach | 4 | registered influence — Meta files Lobbying Disclosure Act reports in 2026 listing artificial intelligence among its issues, with outside registrants also filing for it. | official 2026 |
| METR | entrenchment | 8 | program of record — EU tender EC-CNECT/2025/OP/0032 Lot 3 (loss of control risk modelling) awarded to EquiStamp, METR and Epoch AI for EUR 1,167,484 over 36 months, signed 15 Dec 2025 (TED notice not re-fetched). | official 2025-12-15 |
| METR | entrenchment | 6 | operating — METR operates today; its 14 Aug 2026 funding update reports about $71M raised in six months. | first-hand 2026-08-14 |
| METR | entrenchment | 4 | cited standard — METR's common-elements page says frontier safety policies were 'initially introduced by METR in 2023' and that twelve companies have published them. | first-hand |
| METR | funding insulation | 6 | independent funding — METR (14 Aug 2026): 'We have not accepted funding from these companies, and we do not accept donations made by or at the direction of their staff'; it adds that frontier companies 'provide a significant amount of free tokens'. | first-hand 2026-08-14 |
| METR | reach | 6 | output acted on — METR (Model Evaluation & Threat Research) evaluator-for OpenAI | official |
| METR | reach | 6 | output acted on — METR (Model Evaluation & Threat Research) contract European AI Office | official |
| METR | reach | 6 | output acted on — METR (Model Evaluation & Threat Research) contract Redwood Research | official |
| METR | reach | 6 | output acted on — METR's Claude Opus 5.5 summary: 'We signed off on this final text from the Claude Opus 5.5 system card'; OpenAI's GPT-4 system card credited METR's predecessor ARC for the autonomy evaluation. | first-hand 2026-09-22 |
| METR | reach | 6 | output acted on — The UK AI Security Institute's incident report states it intends to work with METR for an independent third-party review. | official 2026-08-04 |
| METR | reach | 6 | output acted on — METR (Model Evaluation & Threat Research) evaluator-for Anthropic | official |
| MICHAEL AIRD | durability | 6 | built operating — Longview and MATS bios call him co-founder of the Institute for AI Policy and Strategy, which still lists staff in 2026. | first-hand |
| MICHAEL AIRD | insulation | 2 | at will employee — Michael Aird employed-by Longview Philanthropy | official |
| MICHAEL AIRD | insulation | 2 | at will employee — Michael Aird employed-by RAND | official |
| MICHAEL AIRD | leverage | 4 | advisory seat — Michael Aird advisory Centre for the Governance of AI | official |
| MICHAEL AIRD | leverage | 4 | advisory seat — IAPS team page lists Michael Aird under Affiliates with the title Advisor. | first-hand |
| MICHAEL AIRD | leverage | 4 | advisory seat — Longview's bio states Michael leads its AI grantmaking, and Longview recommends grants to the philanthropists it advises; he is also a board member at MATS. | first-hand |
| MICHAEL AIRD | leverage | 4 | advisory seat — Michael Aird board-seat MATS Research | official |
| MICHAEL KRATSIOS | durability | 8 | program of record — Genesis Mission, launched by EO 14363 with the APST leading, is a standing federal AI-for-science program (24 DOE industry agreements of 18 Dec 2025 per the profile; not re-fetched). | official 2025-11-24 |
| MICHAEL KRATSIOS | durability | 4 | cited work — The AI Action Plan 'Winning the AI Race: America's AI Action Plan', released 23 Jul 2025, is the cited federal AI policy document. | official 2025-07-23 |
| MICHAEL KRATSIOS | insulation | 4 | appointed officer — The White House director page calls him 'the thirteenth Director of the White House Office of Science and Technology Policy', a presidential appointee; Senate confirmation 74-25 on 25 Mar 2025 is the profile's figure, vote not re-fetched. | official 2025-03-25 |
| MICHAEL KRATSIOS | leverage | 8 | holds policy pen — EO 14179 tasks the Assistant to the President for Science and Technology, with the AI Czar and NSA, to develop the AI Action Plan; he released it as OSTP Director on 23 Jul 2025 as the federal AI deployment, procurement and export framework. | official 2025-07-23 |
| MICHAEL KRATSIOS | leverage | 8 | holds policy pen — EO 14363 provides that the APST 'shall provide general leadership of the Mission, including coordination of participating executive departments and agencies' and 'the issuance of guidance' for the Genesis Mission. | official 2025-11-24 |
| MICHAEL KRATSIOS | leverage | 8 | holds policy pen — EO 14434 (29 Sep 2026) orders the APST to submit within 60 days proposed legislative language defining 'Super Intelligence' and 'SI'. | official 2026-09-29 |
| MICHAEL KRATSIOS | leverage | 4 | advisory seat — Michael Kratsios board-seat Foundation for American Innovation | official |
| MICHAEL SELLITTO | durability | 5 (ended) | built operating — CNAS bio: he was founding deputy director of Stanford HAI, where he helped build the institute's strategy. | press |
| MICHAEL SELLITTO | durability | 1 | publications — Michael Sellitto co-author Amanda Askell | scholarly |
| MICHAEL SELLITTO | durability | 1 | publications — Michael Sellitto co-author Paul Christiano | scholarly |
| MICHAEL SELLITTO | durability | 1 | publications — Michael Sellitto co-author Jack Clark | scholarly |
| MICHAEL SELLITTO | durability | 1 | publications — Michael Sellitto co-author Jared Kaplan | scholarly |
| MICHAEL SELLITTO | durability | 1 | publications — Michael Sellitto co-author Holden Karnofsky | scholarly |
| MICHAEL SELLITTO | durability | 1 | publications — Michael Sellitto co-author Logan Graham | scholarly |
| MICHAEL SELLITTO | insulation | 2 | at will employee — Michael Sellitto employed-by NSC | official |
| MICHAEL SELLITTO | insulation | 2 | at will employee — Michael Sellitto employed-by Stanford University | official |
| MICHAEL SELLITTO | insulation | 2 | at will employee — Michael Sellitto employed-by State Dept | official |
| MICHAEL SELLITTO | leverage | 4 | advisory seat — Michael Sellitto board-seat Frontier Model Forum | official |
| MICHAEL SHELLENBERGER | durability | 6 | built operating — Public, which he founded and owns, publishes the Twitter Files and other reporting. | first-hand |
| MICHAEL SHELLENBERGER | insulation | 8 | founder control — Public says it has no outside investors and is 'owned entirely by our founder', Michael Shellenberger. | first-hand |
| MICHAEL SHELLENBERGER | leverage | 2 | public voice — Testified to the House Weaponization Subcommittee on March 9 2023, recommending 'Defund the Censorship Industrial Complex'. | official 2023-03-09 |
| MICROSOFT | entrenchment | 8 | program of record — Microsoft reported commercial remaining performance obligation up 84% to $678 billion and over 30 million paid Microsoft 365 Copilot seats in fiscal Q4 2026. | first-hand 2026-07-29 |
| MICROSOFT | reach | 8 | contract or gate — GSA's OneGov agreement with Microsoft makes Microsoft 365 Copilot available at no cost for up to 12 months for Microsoft G5 customers across federal agencies. | official 2025-09-02 |
| MICROSOFT | reach | 4 | convenes or funds — Anthropic funded-by Microsoft | official |
| MICROSOFT | reach | 4 | convenes or funds — Partnership on AI funded-by Microsoft | official |
| MICROSOFT | reach | 4 | convenes or funds — AI Verification and Evaluation Research Institute (AVERI) funded-by Microsoft | official |
| MICROSOFT | reach | 4 | convenes or funds — HiddenLayer funded-by Microsoft | official |
| MIKE BENZ | durability | 6 | built operating — FFO's own about page says Benz founded the Foundation for Freedom Online and is its Executive Director; the organization's site operates today. | first-hand |
| MIKE BENZ | durability | 3 | cited work — NBC News reported that Benz's work has been cited in congressional hearings. | press 2023-10-06 |
| MIKE BENZ | leverage | 2 | public voice — Benz filed a Supreme Court amicus brief in Murthy v. Missouri (No. 23A243) as Executive Director of the Foundation for Freedom Online. | official 2023-09-20 |
| MIKE MCCORMICK | durability | 6 | built operating — Mike McCormick founded Halcyon Futures | official |
| MIKE MCCORMICK | insulation | 4 | appointed officer — ProPublica's extract of Halcyon Futures Inc's Form 990 lists Michael Mccormick as Chief Executive Officer. | official |
| MIKE MCCORMICK | leverage | 4 | advisory seat — Mike McCormick board-seat Seismic Catalyst Foundation | official |
| MIKE MCCORMICK | leverage | 4 | advisory seat — Mike McCormick board-seat Transluce | official |
| MIKE MCCORMICK | leverage | 4 | advisory seat — Mike McCormick board-seat AI Verification and Evaluation Research Institute (AVERI) | official |
| MIKE MCCORMICK | leverage | 4 | funds others — Halcyon Futures, which he founded and leads as CEO, describes itself as a grant-maker, incubator and investment firm and is named on AVERI's own funders list. | first-hand |
| MILES BRUNDAGE | durability | 6 | built operating — Miles Brundage founded AI Verification and Evaluation Research Institute (AVERI) | official |
| MILES BRUNDAGE | durability | 6 | built operating — He co-founded and runs AVERI, a 501(c)(3) launched in January 2026 that is operating, listing funders and staff. | first-hand 2026-01-15 |
| MILES BRUNDAGE | durability | 1 | publications — Miles Brundage co-author Logan Graham | scholarly |
| MILES BRUNDAGE | insulation | 4 | appointed officer — AVERI's team page says he serves as Executive Director, sets strategic direction and leads the team, under a three-member board. | first-hand |
| MILES BRUNDAGE | leverage | 4 | advisory seat — Miles Brundage advisory European AI Office | official |
| MILES BRUNDAGE | leverage | 4 | advisory seat — Miles Brundage advisory Epoch AI | official |
| MILES BRUNDAGE | leverage | 4 | advisory seat — Miles Brundage advisory RAND | official |
| MILES BRUNDAGE | leverage | 4 | advisory seat — Miles Brundage advisory Axon Enterprise | official |
| MACHINE INTELLIGENCE RESEARCH INSTITUTE | entrenchment | 6 | operating — Machine Intelligence Research Institute Inc, a 501(c)(3) since April 2001, reports net assets of $15,242,215 for FY2024. | official 2025-11-07 |
| MACHINE INTELLIGENCE RESEARCH INSTITUTE | funding insulation | 6 | independent funding — MIRI's FY2024 return shows revenue of $1,534,913 (contributions $595,907, investment income $894,015); it sells nothing to labs and grades none. | official 2025-11-07 |
| MACHINE INTELLIGENCE RESEARCH INSTITUTE | reach | 2 | publishes — MIRI publishes research, a draft treaty text and policy papers on stopping superintelligence development. | first-hand |
| MISTRAL AI | entrenchment | 8 | program of record — The French armed forces framework agreement and 125+ enterprise customers constitute Mistral's installed base. | first-hand 2026-09-08 |
| MISTRAL AI | funding insulation | 6 | independent funding — Mistral raised EUR 3 billion in a Series D at over EUR 21 billion post-money, led by Samsung with EQT and PSG as co-leads; it supports 125+ enterprises. | first-hand 2026-09-08 |
| MISTRAL AI | reach | 7 | contract or gate — France's Ministry of the Armed Forces awarded Mistral AI a framework agreement covering the armed forces, directorates, services and affiliated public entities. | press 2026-01-08 |
| MLCOMMONS | entrenchment | 6 | operating — MLCommons Association files as a 501(c)(6) with 2024 revenue of $7,816,648. | official 2024-12-31 |
| MLCOMMONS | entrenchment | 4 | cited standard — MLCommons states two ISO/IEC 42119 standards cite its benchmarks, and the EuroHPC AI Gigafactories tender scores bidders on MLPerf Training v5.1 and Inference v5.1. | first-hand 2025-11-10 |
| MLCOMMONS | funding insulation | 4 | graded funding disclosed — MLCommons publishes dues of $90,000 a year plus a $90,000 initiation fee for large for-profit members and $18,000 plus $18,000 for small ones; the working group that builds AILuminate includes the labs whose systems it grades. | first-hand |
| MLCOMMONS | reach | 8 | contract or gate — MLCommons states the EuroHPC AI Gigafactories tender specifications name MLPerf as the standard for measuring bidders' technical performance, and proposals are scored on MLPerf Training v5.1 and Inference v5.1. | first-hand 2026-10-13 |
| MLCOMMONS | reach | 6 | output acted on — MLCommons states two ISO/IEC 42119 standards on AI testing 'cite MLCommons benchmarks as examples of standardized testing methodologies'. | first-hand 2025-11-10 |
| NAT FRIEDMAN | durability | 6 | built operating — The Vesuvius Challenge site (foundingDate 2023) names him a founder and shows the program operating with prizes awarded. | first-hand 2023-01-01 |
| NAT FRIEDMAN | insulation | 1 | at will employee — Nat Friedman employed-by Meta | press |
| NAT FRIEDMAN | leverage | 4 | funds others — Artificial Intelligence Underwriting Company (AIUC) funded-by Nat Friedman | official |
| NAT FRIEDMAN | leverage | 4 | funds others — Artificial Analysis funded-by Nat Friedman | official |
| NAT FRIEDMAN | leverage | 4 | funds others — Epoch AI funded-by Nat Friedman | official |
| NAT FRIEDMAN | leverage | 4 | funds others — The Vesuvius Challenge, which lists him as a founder alongside Daniel Gross and Brent Seales, states that over $1,958,500 in prizes has been awarded. | first-hand |
| NAT FRIEDMAN | leverage | 4 | funds others — Mechanize funded-by Nat Friedman | official |
| NAT FRIEDMAN | leverage | 3 | funds others — Situational Awareness LP funded-by Nat Friedman | press |
| NATE SOARES | durability | 2 | publications — He is lead author of the MIRI paper Corrigibility (AAAI-15 workshop) and co-author with Yudkowsky of If Anyone Builds It, Everyone Dies (Little, Brown, 16 September 2025). | first-hand 2025-09-16 |
| NATE SOARES | durability | 2 | publications — Nate Soares co-author Eliezer Yudkowsky | official |
| NATE SOARES | insulation | 4 | appointed officer — MIRI's page says he is the President of the Machine Intelligence Research Institute; its team page lists him as President and director. | first-hand |
| NATE SOARES | insulation | 2 | at will employee — Nate Soares employed-by Google / Alphabet | official |
| NATE SOARES | insulation | 2 | at will employee — Nate Soares employed-by Machine Intelligence Research Institute (MIRI) | official |
| NATE SOARES | insulation | 2 | at will employee — Nate Soares employed-by Microsoft | official |
| NATE SOARES | insulation | 2 | at will employee — Nate Soares employed-by National Institute of Standards and Technology | official |
| NATE SOARES | leverage | 4 | advisory seat — Nate Soares advisory Survival and Flourishing Fund | official |
| NEEL NANDA | durability | 6 | built operating — The TransformerLens repository (a library for mechanistic interpretability) is described as created by Neel Nanda and is maintained under the TransformerLensOrg organisation. | first-hand |
| NEEL NANDA | durability | 2 | publications — Neel Nanda co-author Daniel Kokotajlo | official |
| NEEL NANDA | durability | 1 | publications — Neel Nanda co-author Jacob Steinhardt | scholarly |
| NEEL NANDA | durability | 1 | publications — Neel Nanda co-author Lawrence Chan | scholarly |
| NEEL NANDA | insulation | 2 | at will employee — Neel Nanda employed-by DeepMind | official |
| NEEL NANDA | insulation | 2 | at will employee — Neel Nanda employed-by Anthropic | official |
| NEEL NANDA | leverage | 4 | advisory seat — Neel Nanda advisory MATS Research | official |
| NEEL NANDA | leverage | 4 | funds others — His Manifund regrantor page lists him as lead of the mech interp team at Google DeepMind with a regranting budget, and a grant to Forethought that he asked to apply and supports. | first-hand |
| NEURALINK | entrenchment | 6 | operating — ClinicalTrials.gov lists Neuralink Corp as sponsor of the PRIME study (NCT06429735), status Recruiting, start 9 January 2024. | official 2024-01-09 |
| NEURALINK | reach | 4 | registered influence — Arnold & Porter Kaye Scholer LLP lobbying filing lists Neuralink Corp. as client on 'Issues related to Medicare coverage and reimbursement pathways for breakthrough medical devices' and access to neurotechnology/brain-computer interface products. | official 2026-10-19 |
| NEWSGUARD | entrenchment | 6 | operating — NewsGuard's investors page lists thirty outside investors and describes current revenue from subscriptions and data licensing; the company is a plaintiff in NewsGuard Technologies v. FTC (D.D.C. 1:26-cv-00353). | first-hand |
| NEWSGUARD | funding insulation | 5 | independent funding — Co-CEO Crovitz wrote that NewsGuard's Pentagon contract 'is a single-digit percent of our revenues'. | press |
| NEWSGUARD | funding insulation | 4 | graded funding disclosed — NewsGuard's investors page says its revenue comes from consumer subscriptions and from businesses licensing its data, and that co-CEOs Brill and Crovitz hold governing rights and control the editorial products. | first-hand |
| NEWSGUARD | reach | 8 (ended) | contract or gate — USAspending records Air Force contract FA864921P1569, 'MISINFORMATION FINGERPRINTS', $749,387 to NewsGuard Technologies, 7 September 2021 to 8 December 2022. | official 2021-09-07 |
| NEWSGUARD | reach | 6 | output acted on — NewsGuard contract DoD | official |
| NEWSGUARD | reach | 4 | registered influence — ACG Advocacy registered under the Lobbying Disclosure Act with NewsGuard Technologies as client on 'Efforts to address misinformation online. Section 230 Reform.' | official 2021-07-19 |
| NICHOLAS CARLINI | durability | 4 | cited work — ICML 2018 listed Obfuscated Gradients Give a False Sense of Security (Athalye, Carlini, Wagner) among its awards; his homepage says his papers received best paper awards at IEEE S&P, EuroCrypt, USENIX Security and ICML. | official 2018-07-10 |
| NICHOLAS CARLINI | insulation | 2 | at will employee — His homepage lists him as Research Scientist, Anthropic; no term or protection is stated. | first-hand 2025-03-05 |
| NICHOLAS CARLINI | insulation | 2 | at will employee — Nicholas Carlini employed-by DeepMind | official |
| NICHOLAS CARLINI | leverage | 2 | public voice — His homepage lists papers on attacks on machine learning models and states he works at Anthropic studying what bad things you could do with, or do to, language models. | first-hand |
| NICK BOSTROM | durability | 2 | publications — His site lists Superintelligence: Paths, Dangers, Strategies (Oxford University Press, 2014), which remains in print. | first-hand 2014-07-03 |
| NICK BOSTROM | insulation | 0 (ended) | removed not reinstated — The Future of Humanity Institute, of which he was Founding Director, states that Oxford's Faculty of Philosophy declined to renew remaining staff contracts and the Institute was closed down on 16 April 2024. | first-hand 2024-04-16 |
| NICK BOSTROM | leverage | 2 | public voice — His site states he is Founder and Principal Researcher of the nonprofit Macrostrategy Research Initiative and author of Superintelligence (2014). | first-hand |
| NSA MASS-SURVEILLANCE STACK (FISA §702 / PRISM–UPSTREAM / FIVE EYES) | entrenchment | 10 | statute or infrastructure — The 2024 Act (Pub. L. 118-49) reauthorized Section 702; the statute lapsed on 12 June 2026 but existing FISC-certified directives continue in effect under the Title VII transition clause (CRS LSB11444). | official 2024-04-20 |
| NSA MASS-SURVEILLANCE STACK (FISA §702 / PRISM–UPSTREAM / FIVE EYES) | reach | 10 | statutory power — 50 U.S.C. 1881a(i) lets the Attorney General and the DNI direct an electronic communication service provider to provide all information, facilities or assistance necessary for the acquisition, enforceable by FISC order and contempt. | official 2008-07-10 |
| NSO GROUP | entrenchment | 6 | operating — The N.D. Cal. order of 17 October 2025 records NSO's chief executive testifying defendants 'have installation vectors for Pegasus today' and NSO's brief calling Pegasus its flagship product. | official 2025-10-17 |
| NSO GROUP | funding insulation | 4 | graded funding disclosed — NSO's report states its customers are state law-enforcement and intelligence agencies (60 customers in 40 countries); NSO says it licenses 'only to the law enforcement and intelligence agencies of sovereign states'. | first-hand 2021-06-30 |
| NSO GROUP | reach | 8 | contract or gate — NSO's Transparency and Responsibility Report states it licenses Pegasus to sovereign states and state agencies and counts 60 customers in 40 countries. | first-hand 2021-06-30 |
| BEHAVIOURAL INSIGHTS TEAM | entrenchment | 6 | operating — BIT's own blog posts a trial result dated 24 September 2026 and Companies House lists Nesta as its current controlling owner. | first-hand 2026-09-24 |
| BEHAVIOURAL INSIGHTS TEAM | funding insulation | 4 | parent owned — Companies House lists Nesta, the charity, as the person with significant control of the Behavioural Insights Team, with 75% or more of the voting rights. | official |
| BEHAVIOURAL INSIGHTS TEAM | reach | 6 | output acted on — BIT reports that the Government of Kazakhstan 'began scaling the intervention across regions' after a trial run by a behavioural insights unit created with BIT's support. | first-hand 2026-09-24 |
| BEHAVIOURAL INSIGHTS TEAM | reach | 6 | output acted on — Behavioural Insights Team contract UK Government | official |
| OFCOM | entrenchment | 10 | statute or infrastructure — Office of Communications Act 2002 section 1 establishes OFCOM as a body corporate. | official 2002-07-01 |
| OFCOM | funding insulation | 8 | endowment or appropriation — Ofcom's 2024/25 annual report states its sources of income are grant-in-aid, WTA receipts retention, stakeholder income and other income. | official 2025-03-31 |
| OFCOM | reach | 10 | statutory power — Online Safety Act 2023 Schedule 13 sets the maximum Ofcom penalty at the greater of 18 million pounds or ten percent of qualifying worldwide revenue, and section 1 confers the regulator functions on OFCOM. | official 2023-10-26 |
| OMER NEVO | durability | 6 | built operating — He co-founded Irregular (formerly Pattern Labs, 2023), which the paper describes as an AI security lab providing red-teaming, evaluation and auditing services to frontier labs and governments, and which raised $80M led by Sequoia. | first-hand 2025-09-17 |
| OMER NEVO | insulation | 4 | appointed officer — A paper he co-authored states he is employed by Irregular and is a co-founder who maintains equity; the stated role is CTO per his own byline and press; no control is stated. | first-hand 2026-06-23 |
| OMER NEVO | insulation | 1 | at will employee — Omer Nevo employed-by Google / Alphabet | press |
| OMER NEVO | leverage | 6 | runs org acted on — As co-founder and CTO of Irregular, whose cyber evaluation OpenAI's GPT-6 Astra page cites under External Evaluations for Cyber Capabilities (Astra solved 86/226 challenges), and which Google DeepMind's Gemini 4 report says it co-developed a benchmark with. | official 2026-09-01 |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | entrenchment | 6 | operating — Coefficient Giving's about page states it is a current philanthropic funder and advisor directing over $7 billion in grants since 2014. | first-hand |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | funding insulation | 8 | endowment or appropriation — Good Ventures Foundation, which Coefficient Giving serves as outsourced foundation staff and calls its largest funding partner, reports total assets of $10,107,955,038 for the fiscal year ended June 2025 (IRS Form 990-PF data). | official 2025-06-30 |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Coefficient Giving (formerly Open Philanthropy) states it is a philanthropic funder and advisor that has directed over $7 billion in grants since 2014. | first-hand |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Apart Research funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — FAR.AI funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Palisade Research funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Arvind Narayanan funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Redwood Research funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Generality Labs funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Princeton Holistic Agent Leaderboard (HAL) funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Machine Intelligence Research Institute (MIRI) funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — LawZero funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Resolution funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — SecureBio funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Arcadia Impact funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Atlantic Council funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — AI Verification and Evaluation Research Institute (AVERI) funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — Center for Humane Technology funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — MATS Research funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 4 | convenes or funds — SeedAI funded-by Open Philanthropy | official |
| OPEN PHILANTHROPY (COEFFICIENT GIVING) | reach | 3 | convenes or funds — Epoch AI funded-by Open Philanthropy | press |
| OPENAI | entrenchment | 8 | program of record — Microsoft agreement of 28 Oct 2025: IP rights for models and products through 2032, and an incremental $250 billion Azure purchase. | first-hand 2025-10-28 |
| OPENAI | entrenchment | 6 | operating — OpenAI operates as OpenAI Group PBC under the Foundation since the 28 Oct 2025 recapitalization. | first-hand 2025-10-28 |
| OPENAI | funding insulation | 6 | independent funding — OpenAI structure page: 'The OpenAI Foundation continues to control the OpenAI Group'; through special voting rights it 'appoints all members of the board of directors of OpenAI Group and can replace directors at any time'; Foundation holds 26 percent equity, Microsoft about 27 percent. This is a nonprofit's special rights, not majority equity, and the Delaware and California AGs attached non-objection conditions, so 'never overruled' is not established. | first-hand 2025-10-28 |
| OPENAI | reach | 8 | contract or gate — OpenAI for Government: pilot with the Defense Department CDAO, 'with a $200 million ceiling'. | first-hand 2025-06-16 |
| OPENAI | reach | 8 | contract or gate — GSA OneGov: 'Every participating U.S. federal agency will have access for a nominal fee of $1 to ChatGPT Enterprise for one year'. | official 2025-08-06 |
| OPENAI | reach | 4 | convenes or funds — Partnership on AI funded-by OpenAI | official |
| OPENAI | reach | 4 | convenes or funds — Resolution funded-by OpenAI | official |
| OPENAI | reach | 4 | convenes or funds — SeedAI funded-by OpenAI | official |
| OPENAI | reach | 3 | convenes or funds — Epoch AI funded-by OpenAI | press |
| OPENAI FOUNDATION | entrenchment | 6 | operating — OpenAI states the nonprofit is now the OpenAI Foundation, controlling OpenAI Group PBC since the 28 October 2025 recapitalization. | first-hand 2025-10-28 |
| OPENAI FOUNDATION | funding insulation | 8 | endowment or appropriation — OpenAI states the OpenAI Foundation holds a 26% equity stake in OpenAI Group worth approximately $130B plus a warrant, and appoints all members of the Group's board. | first-hand 2025-10-28 |
| OPENAI FOUNDATION | reach | 8 | contract or gate — Delaware Attorney General's statement says the Foundation's Safety and Security Committee remains a committee of the nonprofit with power to require mitigation, up to halting the release of models or AI systems. | official 2025-10-28 |
| OPENAI FOUNDATION | reach | 4 | convenes or funds — SecureBio funded-by OpenAI Foundation | official |
| PALANTIR TECHNOLOGIES | entrenchment | 8 | program of record — Maven Smart System vehicle to May 2029 ($1.275B ceiling) and the Army enterprise agreement of up to 10 years; ICE ICM order to Sep 2027; NHS England contract with extensions to Mar 2031. | official 2024-05-29 |
| PALANTIR TECHNOLOGIES | funding insulation | 8 | endowment or appropriation — 2026 proxy: Class F stock is held in the Founder Voting Trust, set so founders control up to 49.999999 percent of the vote (Karp 11.4, Thiel 7.9, Cohen 3.0); fits 'founder voting control under 50 percent' (proxy not re-fetched, SEC blocked). | first-hand 2026-04-24 |
| PALANTIR TECHNOLOGIES | funding insulation | 6 | independent funding — FY2025 10-K: top three customers 16 and 17 percent of revenue for 2025 and 2024 across 954 customers (government segment 54 percent); Q2 2026 10-Q: no customer over 10 percent. | first-hand 2026-02-01 |
| PALANTIR TECHNOLOGIES | reach | 8 | contract or gate — USAspending: Palantir USG Inc holds the Army Maven Smart System IDV (W911QX24D0012), ceiling $1,275,000,000, performance 29 May 2024 to 28 May 2029. | official 2024-05-29 |
| PALANTIR TECHNOLOGIES | reach | 8 | contract or gate — Army notice (31 Jul 2025): enterprise agreement consolidates 75 contracts into one, up to 10 years, $10 billion cap (Army.mil page not readable here; per the profile). | official 2025-07-31 |
| PALISADE RESEARCH | entrenchment | 6 | operating — ProPublica's IRS filing summary lists Palisade Research Inc, a 501(c)(3), with 2024 revenue of $3,232,257. | official 2024-12-31 |
| PALISADE RESEARCH | entrenchment | 3 | cited standard — Google DeepMind's interpretability team published a write-up investigating the shutdown-resistance environment introduced by Palisade Research. | scholarly 2025-07-14 |
| PALISADE RESEARCH | funding insulation | 6 | independent funding — SFF's 2026 recommendations table lists Palisade Research general support of $1,842,000 funded by Jaan Tallinn and Dustin Moskovitz, and IRS data show 2024 contributions of $2,887,428 of $3,232,257 revenue; no AI-developer payment appears in either record. | official 2026-01-01 |
| PALISADE RESEARCH | reach | 1 | publishes — Palisade authors Schlatter, Weinstein-Raun and Ladish published 'Incomplete Tasks Induce Shutdown Resistance in Some Frontier LLMs' (arXiv 2509.14260; Trans. Mach. Learn. Res. 2026). | scholarly 2025-09-13 |
| PALMER LUCKEY | durability | 7 | program of record — The Army is combining all 120 of its contracts with Anduril into one enterprise agreement with a potential $20 billion ceiling over up to 10 years, centred on Anduril's Lattice (a contract of the company he founded). | press 2026-03-13 |
| PALMER LUCKEY | durability | 6 | built operating — Anduril's own Series H post says When we founded Anduril in 2017 and announces a $5 billion raise at a $61 billion valuation. | first-hand 2026-05-13 |
| PALMER LUCKEY | insulation | 0 (ended) | removed not reinstated — In his TED2025 talk he says he was fired from Facebook after donating $9,000 to a political candidate; Zuckerberg has denied that. | first-hand 2017-03-30 |
| PALMER LUCKEY | leverage | 4 | funds others — FEC itemised receipts filed under Luckey, Palmer with employer Anduril include $200,000 to Trump 47 Committee and $100,000 each to the RNC in the 2024 cycle (source not opened: FEC API rate-limited). | official 2024-06-28 |
| PALMER LUCKEY | leverage | 3 | advisory seat — Palmer Luckey advisory DoD | press |
| PALMER LUCKEY | leverage | 3 | advisory seat — CNBC reports Hegseth named Musk, Luckey and Gingrich as co-leads of Project Meridian, overseen by Pentagon technology chief Emil Michael, with NPR reporting DoD did not disclose whether they are employees or contractors. | press 2026-09-30 |
| PARTNERSHIP ON AI | entrenchment | 6 | operating — PAI's funding page describes a current independent 501(c)(3) funded by philanthropy and corporate entities. | first-hand |
| PARTNERSHIP ON AI | entrenchment | 4 | cited standard — PAI's synthetic-media framework lists 18 named supporters who adopted it and submitted case studies. | first-hand 2023-02-01 |
| PARTNERSHIP ON AI | funding insulation | 4 | graded funding disclosed — PAI's funding page lists eleven corporate funders (Adobe, Amazon, Apple, Google, DeepMind, IBM, Intel, Intuit, Meta, Microsoft, OpenAI) and eight philanthropic funders, with no amounts, and says contributions are non-earmarked. | first-hand |
| PARTNERSHIP ON AI | reach | 6 | output acted on — PAI's Responsible Practices for Synthetic Media names 18 supporters, including Adobe, BBC, Google, Meta, Microsoft, OpenAI and TikTok, each required to submit an in-depth case study; PAI states it will not audit or certify. | first-hand 2023-02-01 |
| PARTNERSHIP ON AI | reach | 4 | convenes or funds — PAI describes itself as a multistakeholder nonprofit of 150+ partner organizations that convenes industry, civil society and academia. | first-hand |
| PATRICIA PASKOV | durability | 4 | cited work — AEF-1, the Forum standard on which she is the ninth of 26 authors, cites her report Preliminary suggestions for rigorous GPAI model evaluations (May 2025) in its references. | first-hand 2025-12-04 |
| PATRICIA PASKOV | insulation | 2 | at will employee — Patricia Paskov employed-by AI Verification and Evaluation Research Institute (AVERI) | official |
| PATRICIA PASKOV | leverage | 4 | advisory seat — Patricia Paskov advisory AI Evaluator Forum | official |
| PATRICIA PASKOV | leverage | 4 | advisory seat — Patricia Paskov advisory MATS Research | official |
| PATRICIA PASKOV | leverage | 4 | advisory seat — Patricia Paskov advisory RAND | official |
| PAUL CHRISTIANO | durability | 6 | built operating — The UK Frontier AI Taskforce report calls him co-founder of ARC, the Alignment Research Centre; ARC's team page shows it operating with him as executive director. | official 2023-09-07 |
| PAUL CHRISTIANO | insulation | 4 | appointed officer — OpenAI's structure page says the OpenAI Foundation appoints the Group board and can replace directors at any time and lists Paul Christiano among the directors; ARC lists him as executive director, board member and President. | first-hand 2026-09-09 |
| PAUL CHRISTIANO | leverage | 8 | signs deployment — OpenAI says Paul Christiano joins the OpenAI Foundation Board and its Safety and Security Committee, which OpenAI says has authority to delay a model release until safety concerns are addressed. | first-hand 2026-09-09 |
| PAUL CHRISTIANO | leverage | 4 | advisory seat — Paul Christiano board-seat OpenAI Foundation | official |
| PAUL CHRISTIANO | leverage | 4 | advisory seat — Paul Christiano advisory UK AI Security Institute | official |
| PAUL NAKASONE | durability | 6 | built operating — Vanderbilt's page lists him as Founding Director of the Vanderbilt Institute of National Security, which operates as an academic national security institute. | first-hand 2024-05-08 |
| PAUL NAKASONE | insulation | 4 | appointed officer — OpenAI's structure page lists Retired U.S. Army General Paul M. Nakasone as a director and says the Foundation appoints the board and can replace directors at any time. | first-hand |
| PAUL NAKASONE | leverage | 8 | signs deployment — OpenAI says its independent Board oversight committee, including retired General Paul Nakasone, oversees model launches and has authority to delay a release until safety concerns are addressed. | first-hand 2024-09-16 |
| PAUL NAKASONE | leverage | 4 | advisory seat — Paul Nakasone board-seat OpenAI Foundation | official |
| PERCY LIANG | durability | 6 | built operating — Marin states it 'began its voyage in 2024 at Stanford University, created by David Hall and Percy Liang' and is now training a 535 billion parameter mixture-of-experts model. | first-hand 2024-01-01 |
| PERCY LIANG | durability | 6 | built operating — Together AI's About page lists Percy Liang with the title Founder. | first-hand |
| PERCY LIANG | durability | 4 | cited work — NIST CAISI's 2025 DeepSeek evaluation cites Cybench (Zhang et al., ending with Dan Boneh, Daniel E. Ho and Percy Liang) and draws tasks from it. | official 2025-09-30 |
| PERCY LIANG | insulation | 4 | appointed officer — CRFM's people page lists Liang as Director of the center; no term or removal protection is stated. | first-hand |
| PERCY LIANG | insulation | 2 | at will employee — Percy Liang employed-by Stanford University | official |
| PERCY LIANG | insulation | 2 | at will employee — Percy Liang employed-by Stanford CRFM | official |
| PERCY LIANG | leverage | 6 | runs org acted on — Stanford CRFM lists 'Percy Liang' with the title 'Director, Computer Science'. | first-hand |
| PERCY LIANG | leverage | 6 | runs org acted on — Stanford CRFM's people page lists Percy Liang as its director, and NIST's CAISI ran Cybench, a benchmark co-authored by Liang, on DeepSeek models in its 2025 evaluation. | official 2025-09-30 |
| PERCY LIANG | leverage | 4 | advisory seat — Percy Liang advisory Transluce | official |
| PETER MATTSON | durability | 6 | built operating — Peter Mattson founded MLCommons | official |
| PETER MATTSON | durability | 1 | publications — Peter Mattson co-author Percy Liang | scholarly |
| PETER MATTSON | insulation | 4 | appointed officer — MLCommons lists Mattson as 'President, Boardmember'; no term or removal protection is stated, and he is also a Google senior staff engineer. | first-hand |
| PETER MATTSON | leverage | 6 | runs org acted on — The EU AI Gigafactories tender scores proposals on MLPerf Training v5.1 and Inference v5.1 benchmarks (MLCommons post, 13 Oct 2026). | first-hand 2026-10-13 |
| PETER MATTSON | leverage | 6 | runs org acted on — MLCommons' leadership page: 'He founded and is President of MLCommons, and founded and was General Chair of the MLPerf consortium that preceded it'; title 'President, Boardmember'. | first-hand |
| PETER THIEL | durability | 6 | built operating — Thiel co-founded Palantir, which still operates and files a proxy statement with him as Chairman. | first-hand 2026-04-23 |
| PETER THIEL | insulation | 8 | founder control — Palantir's proxy: 'Mr. Thiel is one of our co-founders and has served as the Chairman of our Board of Directors since 2003', and he is a founder in the Founder Voting Trust capped at 49.999999% of votes. | first-hand 2026-04-23 |
| PETER THIEL | leverage | 3 | funds others — Clearview AI funded-by Peter Thiel | press |
| PLINY THE LIBERATOR | durability | 6 | built operating — The handle runs OBLITERATUS (abliteration toolkit), L1B3RT4S, CL4R1T4S and LEAKHUB, all linked and live from pliny.gg. | first-hand |
| PLINY THE LIBERATOR | durability | 4 | cited work — pliny.gg lists papers that cite the handle's attacks, including Bijection Learning (arXiv:2410.01294), Jailbreak Paradox (arXiv:2406.12702) and Jailbreak-Zero (arXiv:2601.03265, citing L1B3RT4S); the handle's own page is the source, the papers were not opened. | first-hand |
| PLINY THE LIBERATOR | leverage | 2 | public voice — The handle @elder_plinius publishes jailbreak and leaked-system-prompt repositories (L1B3RT4S, 'TOTALLY HARMLESS LIBERATION PROMPTS FOR GOOD LIL AI'S!', 22.0k stars) and posts as Pliny the Liberator. | first-hand |
| PREAMBLE | entrenchment | 6 | operating — Preamble's home page offers independent AI security assessments, red teaming and patent-backed prompt injection defense; US 12,118,471 is assigned to Preamble Inc (granted 15 October 2024). | official 2024-10-15 |
| PREAMBLE | reach | 6 | output acted on — Preamble contract AFWERX (Department of the Air Force) | official |
| PREAMBLE | reach | 3 | convenes or funds — The HackAPrompt paper thanks its 13 sponsors, Preamble first, 'for their generous donations of funding, credits, and books'. | scholarly 2023-11-24 |
| PREAMBLE | reach | 1 | publishes — McHugh, Sekrst and Cefalu published 'Prompt Injection 2.0: Hybrid AI Threats' (arXiv 2507.13169). | scholarly 2025-07-17 |
| PRINCETON HOLISTIC AGENT LEADERBOARD | entrenchment | 6 | operating — HAL's home page says it has paused updating the leaderboard with new models and is focusing on measuring reliability in AI agents. | first-hand |
| PRINCETON HOLISTIC AGENT LEADERBOARD | funding insulation | 6 | independent funding — HAL states it is funded by Coefficient Giving, Schmidt Sciences, the Princeton AI Lab and two other Princeton initiatives, and thanks OpenAI and Google for API credits to evaluate their models (credit value not published). | first-hand |
| PRINCETON HOLISTIC AGENT LEADERBOARD | reach | 2 | publishes — HAL describes itself as the standardized, cost-aware, and third-party leaderboard for evaluating agents, made by the SAgE team at Princeton University. | first-hand |
| PROMPTFOO | entrenchment | 8 | program of record — OpenAI's announcement describes Promptfoo's tools as trusted by over 25 percent of Fortune 500 companies, plus a widely used open-source CLI and library for evaluating and red-teaming LLM applications (usage figure is the company's own). | official 2026-03-09 |
| PROMPTFOO | funding insulation | 4 | graded funding disclosed — OpenAI, a model vendor whose models Promptfoo's tools test, announced on 9 March 2026 that it is acquiring Promptfoo, so the company's money is now its owner's. | official 2026-03-09 |
| PROMPTFOO | reach | 6 | output acted on — OpenAI announced it is acquiring Promptfoo and states that automated security testing and red-teaming capabilities will become a native part of its Frontier platform. | official 2026-03-09 |
| RAJIV DATTANI | durability | 6 | built operating — Rajiv Dattani co-founder Artificial Intelligence Underwriting Company (AIUC) | official |
| RAJIV DATTANI | durability | 2 | publications — Rajiv Dattani co-author AI Verification and Evaluation Research Institute (AVERI) | official |
| RAJIV DATTANI | insulation | 4 | appointed officer — AIUC's team page says 'Board member @ METR'; METR's about page (not opened) lists him as Advisor and Board Member. No term or removal protection stated. | first-hand |
| RAJIV DATTANI | leverage | 6 | runs org acted on — AIUC's team page lists Rajiv Dattani as one of two co-founders; KPMG LLP became the first Big Four firm to earn AIUC-1 certification (27 Aug 2026). | first-hand 2026-08-27 |
| RAJIV DATTANI | leverage | 4 | advisory seat — Rajiv Dattani board-seat METR (Model Evaluation & Threat Research) | official |
| RAJIV DATTANI | leverage | 4 | advisory seat — Rajiv Dattani advisory Transluce | official |
| RAND CORPORATION | entrenchment | 8 | program of record — The same hearing statement records four standing federally funded R&D centers managed by RAND for DHS and DoD. | official 2023-03-08 |
| RAND CORPORATION | funding insulation | 4 | graded funding disclosed — RAND's audited consolidated financial statements report that federal agencies supplied 89 percent of its US contracts and grants revenue in fiscal 2024 (88 percent in 2023). | first-hand 2024-09-30 |
| RAND CORPORATION | reach | 8 | contract or gate — RAND's president told the Senate Homeland Security committee that RAND currently manages four federally funded research and development centers for the Federal Government, one for DHS and three for DoD. | official 2023-03-08 |
| RAND CORPORATION | reach | 6 | output acted on — RAND contract US Department of Homeland Security | official |
| RAND CORPORATION | reach | 6 | output acted on — RAND contract DoD | official |
| RAND CORPORATION | reach | 4 | registered influence — LDA registration effective 26 June 2025 lists Mercury Public Affairs, LLC as registrant and Rand Corporation as client. | official 2025-06-26 |
| REBECCA FINLAY | durability | 4 | cited work — Under her tenure PAI issued the Guidance for Safe Foundation Model Deployment (24 Oct 2023), developed with consultation from Anthropic, Google DeepMind, Microsoft, OpenAI and others. | first-hand 2023-10-24 |
| REBECCA FINLAY | insulation | 4 | appointed officer — PAI's announcement: she became CEO after an international search, having been on the board since early 2020; the board sets the top official's pay (Form 990 Schedule O, per the page; not opened). | first-hand 2021-10-27 |
| REBECCA FINLAY | insulation | 2 | at will employee — Rebecca Finlay employed-by Canadian Institute for Advanced Research (CIFAR) | official |
| REBECCA FINLAY | leverage | 6 | runs org acted on — Partnership on AI named Rebecca Finlay CEO on 27 Oct 2021; PAI's Oct 2023 Guidance for Safe Foundation Model Deployment carried statements of support from Meta, Microsoft and IBM representatives. | first-hand 2023-10-24 |
| REDWOOD RESEARCH | entrenchment | 6 | operating — Redwood published, jointly with METR, an independent investigation report on the OpenAI / Hugging Face agent incident on its own site. | first-hand 2026-08-26 |
| REDWOOD RESEARCH | entrenchment | 4 | cited standard — The GDM roadmap's reference list cites Redwood's papers (AI control: improving safety despite intentional subversion; Catching AIs red-handed; Ctrl-Z). | first-hand 2026-07-13 |
| REDWOOD RESEARCH | funding insulation | 6 | independent funding — Redwood and METR state that, per standing policy, they took no payment from OpenAI for the independent assessment of the Hugging Face incident, accepting only free API credits (about $400K spent). | first-hand 2026-08-26 |
| REDWOOD RESEARCH | reach | 6 | output acted on — Google DeepMind's AI Control Roadmap (arXiv 2607.13087) builds its untrusted-monitoring and anti-collusion mitigations on Greenblatt, Shlegeris et al. 2024 and thanks Buck Shlegeris and Ryan Greenblatt for feedback on drafts. | first-hand 2026-07-13 |
| REDWOOD RESEARCH | reach | 4 | convenes or funds — Redwood Research advisory Anthropic | official |
| REDWOOD RESEARCH | reach | 4 | convenes or funds — Redwood Research advisory DeepMind | official |
| REDWOOD RESEARCH | reach | 4 | convenes or funds — Redwood Research advisory UK AI Security Institute | official |
| REID HOFFMAN | durability | 6 | built operating — Hoffman's own site and Greylock bio list him as co-founder of LinkedIn and Inflection AI, both operating (pages cited, not opened). | first-hand |
| REID HOFFMAN | insulation | 4 | appointed officer — Microsoft's 8-K: Hoffman has been a Microsoft director since 2017 and will not stand for re-election at the 2026 annual meeting; seat held until then. | first-hand 2026-06-05 |
| REID HOFFMAN | leverage | 4 | advisory seat — Reid Hoffman board-seat Microsoft | official |
| REID HOFFMAN | leverage | 4 | funds others — FEC Schedule A shows Reid Hoffman giving $10,000,000 to Lone Star Rising PAC on 2026-06-30 and $5,000,000 to HMP on 2026-09-10. | official 2026-06-30 |
| REID HOFFMAN | leverage | 4 | funds others — Knight Foundation release (page cited, not opened): Hoffman anchored the 2017 Ethics and Governance of AI Fund with $10M of a $27M fund. | first-hand 2017-01-10 |
| RENÉE DIRESTA | durability | 4 | cited work — The Supreme Court's Murthy v. Missouri opinion names the Election Integrity Partnership as the private entity that alerted Twitter to a flagged post. | official 2024-06-26 |
| RENÉE DIRESTA | durability | 2 | publications — Renee DiResta co-author Kate Starbird | official |
| RENÉE DIRESTA | durability | 1 | publications — Renee DiResta co-author Yoel Roth | scholarly |
| RENÉE DIRESTA | insulation | 2 | at will employee — DiResta is an Associate Research Professor at Georgetown McCourt; no term or tenure is stated. | first-hand |
| RENÉE DIRESTA | leverage | 6 (ended) | runs org acted on — The EIP final report says the SIO team was led by Research Manager Renee DiResta (with Cryst, Thiel, Stamos), and that 35% of URLs the partnership shared with platforms were labeled, removed or soft blocked. | first-hand 2021-03-02 |
| RENÉE DIRESTA | leverage | 2 | public voice — Renee DiResta testified US Senate | official |
| RENÉE DIRESTA | leverage | 2 | public voice — DiResta's own site: Associate Research Professor at Georgetown McCourt; contributing editor at Lawfare. | first-hand |
| RESOLUTION | entrenchment | 6 | operating — Resolution Alignment, Inc. publishes grant, hiring and credits announcements (7 July, 31 August, 13 October 2026) as an operating nonprofit research institute. | first-hand 2026-08-31 |
| RESOLUTION | funding insulation | 6 | independent funding — Resolution states its $160M Coefficient Giving grant ($108M base plus $52M conditional) makes Coefficient its sole funder to start; OpenAI later granted $10M in API credits, bringing total support to $170M (about 6 percent from the lab). | first-hand 2026-07-07 |
| RESOLUTION | reach | 4 | convenes or funds — Resolution announced it is granting $1M to The Formal Commons Project, a two-year project within the Mathlib Initiative. | first-hand 2026-10-13 |
| RILEY GOODSIDE | durability | 4 | cited work — garak's reference documents a goodside probe module: 'Attacks discovered by Riley Goodside' (Tag, Davidjl, ThreatenJSON, WhoIsRiley), used by others to test models. | first-hand |
| RILEY GOODSIDE | insulation | 1 | at will employee — Riley Goodside employed-by DeepMind | press |
| RILEY GOODSIDE | leverage | 1 | public voice — Simon Willison credits Goodside's thread 'Exploiting GPT-3 prompts with malicious inputs that order the model to ignore its previous directions' as the demonstration he named prompt injection. | press 2022-09-12 |
| ROOST (ROBUST OPEN ONLINE SAFETY TOOLS) | entrenchment | 6 | operating — ROOST maintains Osprey, Coop and the ROOST Model Community as an operating 501(c)(3). | first-hand |
| ROOST (ROBUST OPEN ONLINE SAFETY TOOLS) | reach | 6 | output acted on — OpenAI states it worked for months with ROOST to identify developers' critical needs, test gpt-oss-safeguard and produce developer documentation, and that ROOST would establish a model community. | first-hand 2025-10-29 |
| ROOST (ROBUST OPEN ONLINE SAFETY TOOLS) | reach | 6 | output acted on — ROOST's Osprey rules engine page states it is used by Bluesky, Discord, and Matrix. | first-hand |
| RUMMAN CHOWDHURY | durability | 6 | built operating — Rumman Chowdhury founded Humane Intelligence PBC | official |
| RUMMAN CHOWDHURY | insulation | 0 (ended) | removed not reinstated — Her Atlantic account (page cited, not opened): laid off from Twitter's META team in Nov 2022 after the Musk takeover. | first-hand 2022-11-04 |
| RUMMAN CHOWDHURY | leverage | 6 | runs org acted on — Humane Intelligence PBC lists Rumman Chowdhury as Founder & CEO and sells AI test-and-evaluation studies; UNICC announced a partnership with Humane Intelligence (the non-profit) to build UN AI governance tools (22 Sep 2024; that page does not name her). | first-hand |
| RUMMAN CHOWDHURY | leverage | 4 (ended) | advisory seat — Her company page says she served on the inaugural DHS AI Safety and Security Board, which the Federal Register notice terminated effective 7 March 2025. | official 2025-03-07 |
| RUMMAN CHOWDHURY | leverage | 3 | advisory seat — Rumman Chowdhury advisory Stanford CRFM | scholarly |
| RUMMAN CHOWDHURY | leverage | 2 | public voice — Rumman Chowdhury testified U.S. House of Representatives | official |
| RYAN GREENBLATT | durability | 4 | cited work — Google DeepMind's AI Control Roadmap (arXiv 2607.13087, 13 Jul 2026) cites Greenblatt et al., 'AI control: Improving safety despite intentional subversion', and thanks him for feedback. | first-hand 2026-07-13 |
| RYAN GREENBLATT | durability | 1 | publications — Ryan Greenblatt co-author Anthropic | scholarly |
| RYAN GREENBLATT | insulation | 2 | at will employee — Ryan Greenblatt employed-by METR (Model Evaluation & Threat Research) | official |
| RYAN GREENBLATT | insulation | 2 | at will employee — METR's team page lists Ryan Greenblatt as 'Technical Staff'; no term stated. | first-hand |
| RYAN GREENBLATT | leverage | 2 | public voice — First author of 'Alignment faking in large language models' (arXiv 2412.14093) with Anthropic; METR lists him as Technical Staff on investigations and risk assessment. | first-hand 2024-12-18 |
| RYAN KIDD | durability | 6 | built operating — MATS, which Kidd co-founded, is operating (631 researchers, offices in Berkeley and London); he also co-founded LISA. | first-hand |
| RYAN KIDD | durability | 6 | built operating — Ryan Kidd founded MATS Research | official |
| RYAN KIDD | insulation | 4 | appointed officer — MATS FY2024 Form 990 (ProPublica, not opened) lists him as Co-executive Director under an independent three-member board; no protection stated. | first-hand |
| RYAN KIDD | leverage | 8 | signs grants — Kidd's About page: 'As a Manifund Regrantor, I have awarded over $400,000 to AI safety start-ups and projects.' (his own statement; Manifund profile did not render) | first-hand |
| RYAN KIDD | leverage | 6 | runs org acted on — MATS's team page lists Ryan as CEO & Co-Founder; his About page says its programs supported 631 researchers whose alumni co-founded 40+ AI safety organizations. | first-hand |
| RYAN KIDD | leverage | 4 | advisory seat — Ryan Kidd advisory Halcyon Futures | official |
| RYAN KIDD | leverage | 4 | funds others — Apart Research funded-by Ryan Kidd | official |
| RYAN KIDD | leverage | 4 | funds others — Transluce funded-by Ryan Kidd | official |
| SAM ALTMAN | durability | 6 | built operating — Altman co-founded Worldcoin/Tools for Humanity and leads OpenAI, which operates today under the Foundation structure of 28 Oct 2025. | first-hand 2025-10-28 |
| SAM ALTMAN | durability | 4 | cited work — His 16 May 2023 written Senate testimony asked for licensing or registration requirements for AI models above a capability threshold; it is a cited public position. | official 2023-05-16 |
| SAM ALTMAN | insulation | 8 | reinstated — Altman returned as OpenAI CEO within a week of the Nov 2023 removal after the removing board was replaced; OpenAI's structure page now lists him as the CEO member of the Foundation-appointed board. | first-hand 2025-10-28 |
| SAM ALTMAN | insulation | 2 | at will employee — No OpenAI source read states a term or protection for the CEO seat; the Foundation board he sits on 'appoints all members of the board of directors of OpenAI Group and can replace directors at any time'. | first-hand 2025-10-28 |
| SAM ALTMAN | insulation | 0 (reversed) | removed not reinstated — OpenAI's board announced on 17 Nov 2023 that Altman's departure followed a review concluding he 'was not consistently candid in his communications with the board'; the board that wrote it was replaced and he returned as CEO within the week. | first-hand 2023-11-17 |
| SAM ALTMAN | leverage | 8 | signs deployment — OpenAI introduced GPT-6 Astra as 'the world's most intelligent and aligned model' on its own site (Sep 2026); Altman is CEO and a Foundation director. Release authority is inferred from the CEO seat; the page does not name him as signatory. | first-hand 2026-09-04 |
| SAM ALTMAN | leverage | 4 | advisory seat — DHS listed 'Sam Altman, CEO, OpenAI' among the inaugural members of its AI Safety and Security Board. | official 2024-04-26 |
| SAM ALTMAN | leverage | 4 | advisory seat — Sam Altman board-seat OpenAI Foundation | official |
| SAM ALTMAN | leverage | 4 | funds others — FEC itemized receipts list Altman giving $100,000 to Senate Majority PAC (C00484642) on 17 Aug 2022 under employer Y Combinator. | official 2022-08-17 |
| SAM ALTMAN | leverage | 4 | funds others — SMP (Senate Majority PAC, FEC C00484642) funded-by Sam Altman | official |
| SANDER SCHULHOFF | durability | 6 | built operating — Sander Schulhoff founded Learn Prompting | official |
| SANDER SCHULHOFF | durability | 5 | built operating — Sander Schulhoff founded HackAPrompt | scholarly |
| SANDER SCHULHOFF | durability | 3 | cited work — OpenAI's Instruction Hierarchy paper cites Schulhoff et al. 2023 three times as a reference for prompt injection, jailbreaks and prompt extraction. | scholarly 2024-04-19 |
| SANDER SCHULHOFF | durability | 1 | publications — Sander Schulhoff co-author Nicholas Carlini | scholarly |
| SANDER SCHULHOFF | insulation | 2 | at will employee — Sander Schulhoff employed-by University of Maryland | official |
| SANDER SCHULHOFF | leverage | 5 | runs org acted on — Schulhoff is listed as Founder & CEO of Learn Prompting and CEO of HackAPrompt, and OpenAI's Instruction Hierarchy paper cites Schulhoff et al. 2023 (the HackAPrompt dataset paper) for prompt-injection, jailbreak and system-message-extraction attacks. | scholarly 2024-04-19 |
| SANDER SCHULHOFF | leverage | 4 | convenes — Learn Prompting's About page says Schulhoff partnered with OpenAI to run the first AI red-teaming competition, HackAPrompt, which it says now works with frontier AI labs. | first-hand |
| SARAH HECK | insulation | 2 | at will employee — Anthropic's 11 March 2026 announcement names Sarah Heck, who joined as Head of External Affairs, to lead the Public Policy team as Head of Public Policy; no term or protection is stated. | official 2026-03-11 |
| SARAH HECK | leverage | 2 | public voice — Anthropic's July 2025 post quotes Sarah Heck, then Head of Policy, Programs and Partnerships, on the Anthropic Economic Index partnership with the Becker Friedman Institute. | first-hand 2025-07-23 |
| SARAH SCHWETTMANN | durability | 6 | built operating — Sarah Schwettmann founded Transluce | official |
| SARAH SCHWETTMANN | durability | 6 | built operating — Sarah Schwettmann co-founder Jacob Steinhardt | official |
| SARAH SCHWETTMANN | durability | 4 | cited work — OpenAI's GPT-5 system card lists the Transluce o3 truthfulness report, co-authored by Schwettmann, in its references. | official 2025-08-13 |
| SARAH SCHWETTMANN | insulation | 4 | appointed officer — Transluce's FY2024 Form 990 summary lists Sarah Schwettmann as Chief Scientist & Sec/treasurer. | official 2024-12-31 |
| SARAH SCHWETTMANN | leverage | 6 | runs org acted on — OpenAI's GPT-5 system card cites 'Investigating truthfulness in a pre-release o3 model' by Chowdhury, Johnson, Huang, Steinhardt and Schwettmann (Transluce, of which she is co-founder and Chief Scientist) for o3 making false claims about actions it took. | official 2025-08-13 |
| SARAH T. ROBERTS | durability | 6 | built operating — UCLA's faculty directory lists her as faculty director and co-founder of the UCLA Center for Critical Internet Inquiry (C2i2) and co-director of the Minderoo Initiative on Technology & Power. | official |
| SARAH T. ROBERTS | durability | 1 | publications — Behind the Screen, a Yale University Press book by Roberts, remains in print and available. | scholarly 2019-06-25 |
| SARAH T. ROBERTS | leverage | 1 | public voice — Roberts is the author of Behind the Screen: Content Moderation in the Shadows of Social Media (Yale University Press), based on interviews with workers from Silicon Valley to the Philippines. | scholarly 2019-06-25 |
| SATYA NADELLA | durability | 8 | program of record — Microsoft's 28 Oct 2025 restatement with OpenAI: Microsoft IP rights for models and products 'extended through 2032' and a roughly 27 percent stake. | first-hand 2025-10-28 |
| SATYA NADELLA | durability | 6 | built operating — Azure and Copilot installed base (Azure over $100B revenue; 30M+ paid Copilot seats) operates under his CEO tenure since 2014. | first-hand 2026-07-29 |
| SATYA NADELLA | insulation | 4 | appointed officer — Microsoft FY2026 10-K: 'Mr. Nadella was appointed Chairman of the Board in June 2021 and Chief Executive Officer in February 2014'; the proxy records single-class stock with equal votes and under 1 percent beneficial ownership, so no voting control. | first-hand 2026-07-29 |
| SATYA NADELLA | leverage | 6 | runs org acted on — Microsoft FY2026 Q4 release: 'Azure revenue surpassed $100 billion for the first time, and Microsoft 365 Copilot reached over 30 million paid seats'; Nadella is Chairman and CEO. | first-hand 2026-07-29 |
| SATYA NADELLA | leverage | 4 | funds others — Microsoft committed up to $5B in Anthropic and Anthropic committed to buy $30B of Azure; earlier 'multiyear, multibillion dollar' investment in OpenAI. | first-hand 2025-11-18 |
| SAWYER BERNATH | insulation | 4 | appointed officer — FAR.AI's FY2024 Form 990 Part VII lists Sawyer Bernath as Treasurer at 1.00 hour a week with $0 compensation, one of three voting board members. | official 2024-12-31 |
| SAWYER BERNATH | insulation | 2 | at will employee — Sawyer Bernath employed-by Berkeley Existential Risk Initiative (BERI) | official |
| SAWYER BERNATH | insulation | 2 | at will employee — Sawyer Bernath employed-by Tarbell Center for AI Journalism | official |
| SAWYER BERNATH | leverage | 4 | advisory seat — Sawyer Bernath board-seat SecureBio | official |
| SAWYER BERNATH | leverage | 4 | advisory seat — Sawyer Bernath board-seat FAR.AI | official |
| SAWYER BERNATH | leverage | 4 | advisory seat — FAR.AI's team page lists Sawyer Bernath under Board as Board Member (nearest-fit type: a nonprofit board directorship). | official |
| SAYASH KAPOOR | durability | 2 | publications — Sayash Kapoor co-author AI Evaluator Forum | official |
| SAYASH KAPOOR | durability | 2 | publications — Sayash Kapoor co-author Beth Barnes | official |
| SAYASH KAPOOR | durability | 2 | publications — Sayash Kapoor co-author Chris Painter | official |
| SAYASH KAPOOR | durability | 2 | publications — Sayash Kapoor co-author Miles Brundage | official |
| SAYASH KAPOOR | durability | 2 | publications — Sayash Kapoor co-author Arvind Narayanan | official |
| SAYASH KAPOOR | durability | 1 | publications — Sayash Kapoor co-author Percy Liang | scholarly |
| SAYASH KAPOOR | durability | 1 | publications — Sayash Kapoor co-author Stanford CRFM | scholarly |
| SAYASH KAPOOR | durability | 1 | publications — Sayash Kapoor co-author Gillian Hadfield | scholarly |
| SAYASH KAPOOR | durability | 1 | publications — Sayash Kapoor co-author Helen Toner | scholarly |
| SAYASH KAPOOR | durability | 1 | publications — Sayash Kapoor co-author J.J. Allaire | scholarly |
| SAYASH KAPOOR | insulation | 2 | at will employee — Sayash Kapoor employed-by UC Berkeley School of Information | official |
| SAYASH KAPOOR | leverage | 2 | public voice — Kapoor and Narayanan gave written testimony 'Reducing harm from deepfakes' to the Assembly Committee on Science, Innovation, and Technology, dated 4 March 2024, in their personal capacities. | first-hand 2024-03-04 |
| SCALE AI | entrenchment | 8 | program of record — CDAO Production OTA agreement, five-year, ceiling $500M (see above); DIU Thunderforge prototype contract of 5 Mar 2025. | first-hand 2026-05-06 |
| SCALE AI | entrenchment | 6 | operating — Scale AI operates today (CEO Francis deSouza since 10 Aug 2026 per the profile). | first-hand 2026-05-06 |
| SCALE AI | funding insulation | 4 | graded funding disclosed — Scale announced Meta's investment: 'Meta will hold a minority of Scale's outstanding equity'; Meta's 10-Q calls it 'a non-voting minority' with $13.79 billion recorded. Scale is also a data supplier to, and evaluator of, labs. | first-hand 2025-06-12 |
| SCALE AI | reach | 8 | contract or gate — Scale AI: Pentagon CDAO enterprise agreement (Production OTA, awarded Sep 2025) ceiling raised from $100 million to $500 million on 6 May 2026, five-year, usable by any DoD component. | first-hand 2026-05-06 |
| SCALE AI | reach | 6 | output acted on — Scale AI evaluator-for Center for AI Standards and Innovation (NIST) | official |
| SCALE AI | reach | 4 | convenes or funds — SeedAI funded-by Scale AI | official |
| SCHMIDT SCIENCES | entrenchment | 8 | program of record — Schmidt Sciences' AI2050 page describes Eric and Wendy Schmidt's $125 million commitment over five years to support individuals working on AI. | first-hand |
| SCHMIDT SCIENCES | funding insulation | 8 | endowment or appropriation — Form 990-PF for the Eric and Wendy Schmidt Fund for Strategic Innovation (the parent) reports net assets of $1,440,351,492 and charitable disbursements of $315,860,401 for fiscal 2024. | official 2024-12-31 |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Percy Liang funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Stanford CRFM funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Epoch AI funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Adam Gleave funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Gillian Hadfield funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Princeton Holistic Agent Leaderboard (HAL) funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — LawZero funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — METR (Model Evaluation & Threat Research) funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Zico Kolter funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Frontier Model Forum AI Safety Fund funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Dan Hendrycks funded-by Schmidt Sciences | official |
| SCHMIDT SCIENCES | reach | 4 | convenes or funds — Arvind Narayanan funded-by Schmidt Sciences | official |
| SEÁN Ó HÉIGEARTAIGH | durability | 6 | built operating — The University of Cambridge reports the Leverhulme Trust's 10 million pound grant for the Centre for the Future of Intelligence was awarded for a proposal developed with CSER's Executive Director Dr Seán Ó hÉigeartaigh. | official 2015-12-03 |
| SEÁN Ó HÉIGEARTAIGH | durability | 6 | built operating — CSER's staff page says he was the founding Executive Director of CSER (2013-2015) and led its major research projects 2015-2020; CSER operates today. | first-hand |
| SEÁN Ó HÉIGEARTAIGH | insulation | 4 | appointed officer — GovAI's 2025 annual report lists Seán Ó hÉigeartaigh in its Board section. | official |
| SEÁN Ó HÉIGEARTAIGH | insulation | 4 | appointed officer — Seán Ó hÉigeartaigh appointed-by European Commission | official |
| SEÁN Ó HÉIGEARTAIGH | insulation | 2 | at will employee — Seán Ó hÉigeartaigh employed-by University of Cambridge | official |
| SEÁN Ó HÉIGEARTAIGH | insulation | 2 | at will employee — Seán Ó hÉigeartaigh employed-by Future of Humanity Institute (Oxford) | official |
| SEÁN Ó HÉIGEARTAIGH | insulation | 2 | at will employee — Seán Ó hÉigeartaigh employed-by Centre for the Study of Existential Risk (Cambridge) | official |
| SEÁN Ó HÉIGEARTAIGH | insulation | 2 | at will employee — Seán Ó hÉigeartaigh employed-by Leverhulme Centre for the Future of Intelligence (Cambridge) | official |
| SEÁN Ó HÉIGEARTAIGH | leverage | 4 | advisory seat — Seán Ó hÉigeartaigh advisory Survival and Flourishing Corp (SFC) | official |
| SEÁN Ó HÉIGEARTAIGH | leverage | 4 | advisory seat — Seán Ó hÉigeartaigh advisory Survival and Flourishing Fund | official |
| SEÁN Ó HÉIGEARTAIGH | leverage | 4 | advisory seat — Seán Ó hÉigeartaigh board-seat Centre for the Governance of AI | official |
| SEÁN Ó HÉIGEARTAIGH | leverage | 4 | advisory seat — The Leverhulme Centre for the Future of Intelligence reports Professor Seán Ó hÉigeartaigh joined the European Commission's AI Act Scientific Panel for a 24-month renewable term, in personal capacity; the Panel advises the AI Office and national authorities. | first-hand 2026-07-02 |
| SECUREBIO | entrenchment | 6 | operating — SecureBio published its GPT-6 Astra pre-release assessment on 14 September 2026. | first-hand 2026-09-14 |
| SECUREBIO | entrenchment | 4 | cited standard — OpenAI's Astra page reports model results on SecureBio's Virology Capabilities Test and VCT-v2 by name. | first-hand |
| SECUREBIO | funding insulation | 6 | independent funding — SecureBio's assessment states SecureBio AI is not funded by OpenAI, its employees or competitors beyond payment for the assessment at cost, and is primarily funded by Coefficient Giving, the Survival and Flourishing Fund, and Sentinel Bio. | first-hand 2026-09-14 |
| SECUREBIO | reach | 6 | output acted on — SecureBio evaluator-for Meta | official |
| SECUREBIO | reach | 6 | output acted on — SecureBio evaluator-for OpenAI | official |
| SECUREBIO | reach | 6 | output acted on — OpenAI's GPT-6 Astra deployment-safety page carries external biological-capability and bio-safeguard evaluations by SecureBio and reports Astra's highest-observed scores on SecureBio's VCT (57.8 percent) and VCT-v2 (57.7 percent). | first-hand |
| SEEDAI | entrenchment | 6 | operating — SeedAI runs the ongoing AI Primers series and lists current projects on its own site. | first-hand |
| SEEDAI | reach | 4 | convenes or funds — AI Village's announcement names SeedAI, with Humane Intelligence and the AI Vulnerability Database, as a partner hosting the first public generative AI red team at DEF CON 31. | first-hand 2023-05-03 |
| SEEDAI | reach | 4 | convenes or funds — SeedAI's AI Primers are an event series designed to bolster Congressional understanding of AI, each a panel of policy experts and industry leaders moderated by a SeedAI staff member. | first-hand |
| SEISMIC CATALYST FOUNDATION | entrenchment | 6 | operating — ProPublica's IRS data show Seismic Catalyst Foundation, exempt August 2025, reported $5,248,744 revenue (100 percent contributions) and $2,781,299 expenses for calendar 2025. | official 2025-12-31 |
| SEISMIC CATALYST FOUNDATION | reach | 2 | publishes — Seismic's report On the Razor's Edge states it polled 10,000 people across the U.S., U.K., France, Germany and Poland. | first-hand 2025 |
| SENTINEL BIO | entrenchment | 6 | operating — ProPublica's IRS data show Sentinel Bio Inc's first Form 990 (fiscal 2024) with revenue of $10,595,173 and net assets of $9,060,916. | official 2024-12-31 |
| SENTINEL BIO | funding insulation | 6 | independent funding — Sentinel's organization page states its funders are individuals, foundations and donor-advised funds, all funding is unrestricted, funders have no direct influence on its grants, and no donor makes up over 25 percent of funding. | first-hand |
| SENTINEL BIO | reach | 4 | convenes or funds — SecureBio funded-by Sentinel Bio | official |
| SENTINEL BIO | reach | 4 | convenes or funds — BioTrust funded-by Sentinel Bio | official |
| SENTINEL BIO | reach | 4 | convenes or funds — Epoch AI funded-by Sentinel Bio | official |
| SENTINEL BIO | reach | 4 | convenes or funds — RAND funded-by Sentinel Bio | official |
| SENTINEL BIO | reach | 4 | convenes or funds — Active Site funded-by Sentinel Bio | official |
| SENTINEL BIO | reach | 4 | convenes or funds — Engineering Biology Research Consortium (EBRC) funded-by Sentinel Bio | official |
| SENTINEL BIO | reach | 4 | convenes or funds — Johns Hopkins Center for Health Security funded-by Sentinel Bio | official |
| SHYAM SANKAR | durability | 6 | built operating — His own site says he envisaged the role of the Forward Deployed Engineer, pioneering Palantir's definitional engineering model. | first-hand |
| SHYAM SANKAR | insulation | 4 | appointed officer — Palantir's 2026 proxy statement lists Shyam Sankar, 44, as Chief Technology Officer and Executive Vice President, an executive officer, with 2025 total compensation of $1,209,250. | official 2026-04-23 |
| SHYAM SANKAR | leverage | 6 | runs org acted on — The US Army awarded Palantir (whose CTO and EVP is Sankar, an executive officer per the 2026 proxy) an Enterprise Agreement consolidating 75 contracts, with a performance period of up to 10 years and a $10 billion cap. | official 2025-07-31 |
| SHYAM SANKAR | leverage | 4 | advisory seat — Shyam Sankar board-seat Ginkgo Bioworks | official |
| SHYAM SANKAR | leverage | 4 | advisory seat — The Army's release names Shyam Sankar, Chief Technology Officer for Palantir, one of four new Army Reserve Lt. Cols. in Detachment 201, which recruits tech executives to serve part-time as senior advisors. | official 2025-06-13 |
| SHYAM SANKAR | leverage | 4 | advisory seat — Shyam Sankar board-seat Hudson Institute | official |
| SHYAM SANKAR | leverage | 2 | public voice — Shyam Sankar testified US Senate | official |
| SIGNATURE SCIENCE | entrenchment | 8 | program of record — Signature Science's contract-vehicles page lists it as prime or subcontractor on federal vehicles running to 2027-2030, including CTRIC IV (8/14/30), DTRA Combatting WMD RDT&E (6/30/29) and Seaport-NxG (11/30/28). | first-hand |
| SIGNATURE SCIENCE | entrenchment | 6 | operating — Signature Science announced on 21 October 2025 that it was selected for NIST CAISI's Basic Ordering Agreement for chemical and biological AI benchmarks (task orders under a $3M ceiling). | first-hand 2025-10-21 |
| SIGNATURE SCIENCE | funding insulation | 4 | parent owned — Signature Science's own site states it is a subsidiary of Southwest Research Institute, San Antonio, TX. | first-hand |
| SIGNATURE SCIENCE | reach | 6 | output acted on — Signature Science evaluator-for Anthropic | official |
| SIGNATURE SCIENCE | reach | 6 | output acted on — Signature Science contract Center for AI Standards and Innovation (NIST) | official |
| SRIRAM KRISHNAN | durability | 4 (ended) | cited work — His own site says that in the White House role he worked on the AI Action Plan, which he cited among his accomplishments on leaving. | first-hand 2025-07-23 |
| SRIRAM KRISHNAN | insulation | 2 | at will employee — Sriram Krishnan employed-by Meta | official |
| SRIRAM KRISHNAN | insulation | 2 | at will employee — Sriram Krishnan employed-by Microsoft | official |
| SRIRAM KRISHNAN | insulation | 2 | at will employee — Sriram Krishnan employed-by X Corp (formerly Twitter) | official |
| SRIRAM KRISHNAN | leverage | 4 (ended) | advisory seat — His own site says he was Senior Policy Advisor for AI at the White House from January 2025 to June 2026 (nearest-fit type: a White House policy adviser seat). | first-hand 2026-06-30 |
| SRIRAM KRISHNAN | leverage | 4 | funds others — Scale AI funded-by Sriram Krishnan | official |
| STANFORD CENTER FOR RESEARCH ON FOUNDATION MODELS (CRFM) | entrenchment | 6 | operating — CRFM operates as an initiative of Stanford HAI with a published people page, support page and active research and policy output. | first-hand |
| STANFORD CENTER FOR RESEARCH ON FOUNDATION MODELS (CRFM) | reach | 6 | output acted on — US AISI tested o1 on Cybench, 'a publicly available cybersecurity benchmark for language models developed by researchers at Stanford University', in the US-UK joint pre-deployment test. | official 2024-12-18 |
| STANFORD CENTER FOR RESEARCH ON FOUNDATION MODELS (CRFM) | reach | 6 | output acted on — CAISI evaluated DeepSeek models and reference models on Cybench, 'a publicly available cyber benchmark developed by researchers at Stanford University'. | official 2025-09-30 |
| STANFORD INTERNET OBSERVATORY | entrenchment | 2 (ended) | reports persist — The Election Integrity Partnership final report, The Long Fuse, co-produced by SIO, remains available in the Stanford Digital Repository. | first-hand |
| STANFORD INTERNET OBSERVATORY | funding insulation | 6 (ended) | independent funding — The Election Integrity Partnership report thanks SIO's operational funders, Craig Newmark Philanthropies and the William and Flora Hewlett Foundation (a $5 million Newmark gift launched it in 2019). | first-hand |
| STANFORD INTERNET OBSERVATORY | reach | 6 (ended) | output acted on — Meta's December 2021 adversarial threat report says it improved its takedown-archive platform in response to feedback from teams at the Stanford Internet Observatory and others. | official |
| STANFORD INTERNET OBSERVATORY | reach | 5 (ended) | output acted on — TechCrunch reports Re-LAION-5B was re-released with fixes implemented with recommendations from the IWF, Human Rights Watch, the Canadian Center for Child Protection and the now-defunct Stanford Internet Observatory. | press 2024-08-30 |
| STUART RUSSELL | durability | 6 | built operating — The Kavli Center's faculty page lists Stuart Russell as Director of the Center for Human-Compatible AI, launched in 2016 under his lead. | first-hand |
| STUART RUSSELL | durability | 4 | cited work — The Artificial Intelligence: A Modern Approach site (Russell and Norvig) says the textbook is adopted by over 1500 schools. | first-hand |
| STUART RUSSELL | leverage | 4 | convenes — The IASEAI'26 call for papers lists Stuart Russell, President pro tem, on the organizing committee of the association's Paris conference. | first-hand 2026-02-24 |
| SUCHIR BALAJI | durability | 2 (ended) | publications — His fair-use essay remains available on his personal site. | first-hand 2024-10-23 |
| SUCHIR BALAJI | insulation | 2 | at will employee — Suchir Balaji employed-by OpenAI | official |
| SUCHIR BALAJI | leverage | 2 (ended) | public voice — Balaji's own site carries his essay 'When does generative AI qualify for fair use?' dated 23 October 2024, concluding none of the four factors favours ChatGPT's training use. | first-hand 2024-10-23 |
| SUNDAR PICHAI | durability | 6 | built operating — Google's April 2023 announcement, signed by Pichai as CEO, combined the Brain team and DeepMind into Google DeepMind under Demis Hassabis. | first-hand 2023-04-20 |
| SUNDAR PICHAI | insulation | 4 | appointed officer — The same proxy's beneficial-ownership table shows Larry Page at 27.4 and Sergey Brin at 25.3 percent of voting power and Pichai with 227,560 Class A shares and no Class B, so the CEO holds no controlling vote. | official 2026-04-24 |
| SUNDAR PICHAI | leverage | 6 | runs org acted on — Alphabet's 2026 proxy statement names him Chief Executive Officer of Google (since October 2015) and of Alphabet (since December 2019) and a director. | official 2026-04-24 |
| SURVIVAL AND FLOURISHING FUND | entrenchment | 6 | operating — SFF runs annual S-Process grant-recommendation rounds, most recently SFF-2026 recommending $63.13MM. | first-hand |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Epoch AI funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — SecureBio funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — SFF states it has organized ~$216MM in philanthropic gifts and grants through its S-Process recommendation rounds (about $64MM in 2026). | first-hand |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Apart Research funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Apollo Research funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — MATS Research funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Center for Humane Technology funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Centre for the Governance of AI funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — METR (Model Evaluation & Threat Research) funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Machine Intelligence Research Institute (MIRI) funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — AI Impacts funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Center for AI Safety funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — FAR.AI funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Palisade Research funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Seismic Catalyst Foundation funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Resolution funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Collective Intelligence Project funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Alignment Research Center funded-by Survival and Flourishing Fund | official |
| SURVIVAL AND FLOURISHING FUND | reach | 4 | convenes or funds — Redwood Research funded-by Survival and Flourishing Fund | official |
| TAMAY BESIROGLU | durability | 6 | built operating — Epoch's own bio states he co-founded Epoch AI, which still publishes and operates. | first-hand |
| TAMAY BESIROGLU | insulation | 2 | at will employee — Tamay Besiroglu employed-by DeepMind | official |
| TAMAY BESIROGLU | insulation | 1 | at will employee — Business Insider reports Mechanize's co-founder and ex-CEO now works as a research scientist at Google DeepMind, per his LinkedIn profile. | press 2026-09-11 |
| TAMAY BESIROGLU | leverage | 5 (ended) | runs org acted on — TechCrunch named him associate director and co-founder of Epoch AI when he defended its OpenAI-funded FrontierMath benchmark; he has since left to co-lead Mechanize. | press 2025-01-19 |
| TAMAY BESIROGLU | leverage | 4 | advisory seat — Epoch's own bio says he co-founded Epoch AI and remains contributing to the organization as a research advisor. | first-hand |
| TASHA MCCAULEY | durability | 1 | publications — Tasha McCauley co-author Helen Toner | press |
| TASHA MCCAULEY | insulation | 4 | appointed officer — The Form 990 lists her as board Chair (beginning June) of Centre For The Governance Of Ai Inc at $0 compensation; no control or removal protection is stated. | official 2025-11-13 |
| TASHA MCCAULEY | insulation | 1 | at will employee — Tasha McCauley employed-by RAND | press |
| TASHA MCCAULEY | insulation | 0 (ended) | removed not reinstated — OpenAI's 29 November 2023 message names a new initial board of Bret Taylor, Larry Summers and Adam D'Angelo after Altman's return, thanking McCauley and Toner for their work, so her independent-director seat ended and was not restored. | official 2023-11-29 |
| TASHA MCCAULEY | leverage | 4 | advisory seat — Tasha McCauley board-seat Effective Ventures Foundation (UK) | official |
| TASHA MCCAULEY | leverage | 4 | advisory seat — Tasha McCauley board-seat OpenAI | official |
| TASHA MCCAULEY | leverage | 4 | advisory seat — The Form 990 for the year ending December 2024 lists Tasha Mccauley as Chair of the board of Centre For The Governance Of Ai Inc. | official 2025-11-13 |
| TASHA MCCAULEY | leverage | 4 | advisory seat — Tasha McCauley board-seat Centre for the Governance of AI | official |
| THIERRY BRETON | durability | 7 (ended) | program of record — US Under Secretary Sarah Rogers described him as the 'mastermind' of the EU's Digital Services Act, a law that remains in force and is enforced without him. | press 2025-12-24 |
| THIERRY BRETON | insulation | 8 (ended) | statutory term — The Commission's own commissioner page lists him as Commissioner for the Internal Market in the 2019-2024 College; he resigned on 16 September 2024. | official 2024-09-16 |
| THIERRY BRETON | leverage | 8 (ended) | signs enforcement — As Commissioner for the Internal Market the Commission opened its first formal proceedings under the Digital Services Act, against X, on 18 December 2023. | official 2023-12-18 |
| THIERRY BRETON | leverage | 3 | advisory seat — Breton confirmed to AFP he would again be a member of Bank of America's global advisory council, which he called consultative, three meetings a year. | press 2025-01-16 |
| MARIANO-FLORENTINO CUELLAR | durability | 2 | publications — He is named as an author of the California Report on Frontier AI Policy published for the Governor's office. | official 2025-06-17 |
| MARIANO-FLORENTINO CUELLAR | insulation | 4 | appointed officer — Anthropic announced him as its first Chief Global Affairs Officer, leading policy, international engagement and government relations worldwide. | official 2026-08-04 |
| MARIANO-FLORENTINO CUELLAR | insulation | 2 | at will employee — Mariano-Florentino Cuellar employed-by Stanford University | official |
| MARIANO-FLORENTINO CUELLAR | leverage | 4 | advisory seat — Mariano-Florentino Cuellar advisory White House | official |
| MARIANO-FLORENTINO CUELLAR | leverage | 4 | advisory seat — Mariano-Florentino Cuellar board-seat William and Flora Hewlett Foundation | official |
| MARIANO-FLORENTINO CUELLAR | leverage | 4 | advisory seat — Mariano-Florentino Cuellar advisory State Dept | official |
| MARIANO-FLORENTINO CUELLAR | leverage | 4 | advisory seat — Mariano-Florentino Cuellar board-seat Harvard | official |
| MARIANO-FLORENTINO CUELLAR | leverage | 4 (ended) | advisory seat — Anthropic says he served as a Trustee of the Long-Term Benefit Trust since January 2026 and has stepped down to join the company. | official 2026-08-04 |
| MARIANO-FLORENTINO CUELLAR | leverage | 4 | advisory seat — Governor Newsom's 17 June 2025 release names him among the authors of the California Report on Frontier AI Policy. | official 2025-06-17 |
| TOBY ORD | durability | 6 | built operating — Toby Ord co-founder Will MacAskill | official |
| TOBY ORD | durability | 6 | built operating — Toby Ord founded Giving What We Can | official |
| TOBY ORD | durability | 4 | cited work — The UK Prime Minister quoted his words in a UN General Assembly speech. | official 2021-09-22 |
| TOBY ORD | insulation | 4 | appointed officer — The Form 990 lists him as a Director of Centre For The Governance Of Ai Inc; no term or protection is stated. | official 2025-11-13 |
| TOBY ORD | insulation | 2 | at will employee — Toby Ord employed-by Oxford Martin AI Governance Initiative | official |
| TOBY ORD | leverage | 4 | advisory seat — Toby Ord advisory Organisation for Economic Co-operation and Development (OECD) | official |
| TOBY ORD | leverage | 4 | advisory seat — Toby Ord advisory UK Government | official |
| TOBY ORD | leverage | 4 | advisory seat — Toby Ord advisory UN | official |
| TOBY ORD | leverage | 4 | advisory seat — Toby Ord advisory World Health Organization | official |
| TOBY ORD | leverage | 4 | advisory seat — Toby Ord board-seat Centre for the Governance of AI | official |
| TOBY ORD | leverage | 4 | advisory seat — The Oxford Martin AIGI page says he serves on the board of the Centre for the Governance of AI, and the Form 990 lists him as a Director (beginning June) at $0. | official 2025-11-13 |
| TOBY ORD | leverage | 4 | advisory seat — Toby Ord advisory DeepMind | official |
| TOBY ORD | leverage | 4 | advisory seat — Toby Ord advisory WEF | official |
| TOBY ORD | leverage | 2 | public voice — Prime Minister Boris Johnson quoted him by name at the UN General Assembly on 22 September 2021. | official 2021-09-22 |
| TRAIL OF BITS | entrenchment | 8 | program of record — USAspending lists multi-year US government contracts to TRAIL OF BITS INC, including DARPA SIEVE (HR001120C0084, $7,068,272, April 2020 to April 2024) and a DoD SBIR Phase II (W5170125CA190, $1,999,742.83, July 2025 to January 2027). | official |
| TRAIL OF BITS | reach | 6 | output acted on — The audited project's own recap says Trail of Bits submitted 27 private repository advisories and 3 hardening pull requests to OpenClaw, with 23 of 24 severity-rated reports classified as confirmed vulnerabilities. | first-hand 2026-09-21 |
| TRAIL OF BITS | reach | 6 | output acted on — Trail of Bits contract UK AI Security Institute | official |
| TRAJECTORY LABS | entrenchment | 6 | operating — Trajectory Labs, PBC operates a live site listing its founding team, lab customers and AI safety evals, red teaming and RL environment work. | first-hand |
| TRAJECTORY LABS | reach | 6 | output acted on — Trajectory Labs evaluator-for Meta | official |
| TRAJECTORY LABS | reach | 6 | output acted on — Anthropic: 'We commissioned an evaluation from a third party, Trajectory Labs, who tested different models within the latest publicly available versions of Claude Code and Codex', and reports the result in its auto-mode launch post. | first-hand |
| TRAJECTORY LABS | reach | 6 | output acted on — Trajectory Labs evaluator-for DeepMind | official |
| TRAJECTORY LABS | reach | 6 | output acted on — Trajectory Labs evaluator-for Anthropic | official |
| TRANSLUCE | entrenchment | 6 | operating — Transluce operates as a 501(c)(3) (registered name Clarity AI Research Inc.) publishing evaluations and the open-source Docent tool. | first-hand |
| TRANSLUCE | funding insulation | 6 | independent funding — Transluce's policy states that in fiscal 2025 it received 6% of revenue from personal holdings of OpenAI employees and 32% from Anthropic employees, 'no revenue directly from AI developers as organizations', and has carried out no paid evaluations. | first-hand 2026-08-27 |
| TRANSLUCE | reach | 6 | output acted on — OpenAI's GPT-5 system card cites Transluce's o3 truthfulness report (reference 3) in its section on deception. | first-hand 2025-08-13 |
| TRANSLUCE | reach | 6 | output acted on — Transluce contract European AI Office | official |
| TRANSLUCE | reach | 6 | output acted on — Transluce evaluator-for OpenAI | official |
| UK AI SAFETY INSTITUTE | entrenchment | 8 | program of record — The DeepMind MoU runs 36 months; AISI says the government plans 'to move the organisation onto a statutory footing' (a plan, not enacted). | official 2024-11-20 |
| UK AI SAFETY INSTITUTE | entrenchment | 6 | operating — AISI operates within DSIT with pre-deployment access to leading models and £66m per financial year. | official |
| UK AI SAFETY INSTITUTE | funding insulation | 8 | endowment or appropriation — AISI About page: '£66m in funding per financial year and long-term resourcing commitments' as a research organisation within DSIT; government appropriation, not statutory. | official |
| UK AI SAFETY INSTITUTE | reach | 6 | output acted on — UK AI Security Institute evaluator-for OpenAI | official |
| UK AI SAFETY INSTITUTE | reach | 6 | output acted on — UK AI Security Institute evaluator-for DeepMind | official |
| UK AI SAFETY INSTITUTE | reach | 6 | output acted on — US and UK AISI tested the upgraded Claude 3.5 Sonnet 'during a limited period of pre-deployment access' and the 'findings were shared with Anthropic before the model was publicly released'. | official 2024-11-19 |
| UK AI SAFETY INSTITUTE | reach | 6 | output acted on — UK AI Security Institute evaluator-for Anthropic | official |
| UK AI SAFETY INSTITUTE | reach | 6 | output acted on — DSIT-Google DeepMind MoU (11 Dec 2025, 36 months, voluntary, not legally binding): DeepMind provides AISI 'priority technical access to its frontier models'. | official 2025-12-11 |
| UK AI SAFETY INSTITUTE | reach | 4 | convenes or funds — METR (Model Evaluation & Threat Research) funded-by UK AI Security Institute | official |
| UK AI SAFETY INSTITUTE | reach | 4 | convenes or funds — Arcadia Impact funded-by UK AI Security Institute | official |
| UK NATIONAL CYBER SECURITY CENTRE | entrenchment | 6 | operating — The NCSC states it has operated since 2016 as a part of GCHQ. | first-hand |
| UK NATIONAL CYBER SECURITY CENTRE | funding insulation | 8 | endowment or appropriation — The NCSC describes itself as 'a part of GCHQ', and GCHQ is funded through the Single Intelligence Account, the funding vehicle for the Security and Intelligence Agencies (NCSC is not itemised separately). | official 2025-10-27 |
| UK NATIONAL CYBER SECURITY CENTRE | reach | 4 | convenes or funds — The NCSC and CISA published the Guidelines for secure AI system development jointly with partner agencies from many countries and listed contributing organisations including Anthropic, Google DeepMind, Microsoft and OpenAI. | official 2023-11-27 |
| UK ONLINE SAFETY ACT | entrenchment | 10 | statute or infrastructure — The Online Safety Act 2023 (c. 50) is a UK Public General Act that gives Ofcom its online safety functions, including penalties (Schedule 13) and fees (Part 6). | official 2023-10-26 |
| UK ONLINE SAFETY ACT | funding insulation | 4 | graded funding disclosed — Section 84 lets Ofcom charge regulated-service providers fees, and section 88(2)(a) requires that the aggregate fees be sufficient to meet, but not exceed, Ofcom's annual cost of exercising its online safety functions. | official 2024-01-10 |
| UK ONLINE SAFETY ACT | reach | 10 | statutory power — Schedule 13 paragraph 4 of the Online Safety Act 2023 sets Ofcom's maximum penalty on a provider of a regulated service at the greater of GBP 18 million or 10% of qualifying worldwide revenue. | official 2023-10-26 |
| CENTER FOR AI STANDARDS AND INNOVATION (CAISI) | entrenchment | 6 | operating — NIST's CAISSI page (formerly CAISI) lists its mandate to establish voluntary agreements with developers and lead unclassified evaluations, and its completed evaluations. | official |
| CENTER FOR AI STANDARDS AND INNOVATION (CAISI) | funding insulation | 8 | endowment or appropriation — The joint explanatory statement to the Commerce, Justice, Science appropriations provides no less than $55,000,000 for NIST AI research and 'up to $10,000,000' to expand NIST's AI efforts through the U.S. Center for AI Standards and Innovation. | official |
| CENTER FOR AI STANDARDS AND INNOVATION (CAISI) | reach | 6 | output acted on — Center for AI Standards and Innovation (NIST) evaluator-for OpenAI | official |
| CENTER FOR AI STANDARDS AND INNOVATION (CAISI) | reach | 6 | output acted on — OpenAI says it was 'among the first companies to enter into voluntary agreements' with CAISI and that its July 2025 collaboration with CAISI on red-teaming agentic systems fed vulnerability fixes for ChatGPT Agent. | first-hand 2025-09-12 |
| CENTER FOR AI STANDARDS AND INNOVATION (CAISI) | reach | 6 | output acted on — Center for AI Standards and Innovation (NIST) evaluator-for DeepMind | official |
| CENTER FOR AI STANDARDS AND INNOVATION (CAISI) | reach | 6 | output acted on — Center for AI Standards and Innovation (NIST) evaluator-for Anthropic | official |
| VALS AI | entrenchment | 6 | operating — Vals AI operates public benchmarks and leaderboards (Finance Agent, Vals Index) and states its benchmarks 'have rapidly become industry standards'. | first-hand |
| VALS AI | reach | 6 | output acted on — Vals AI evaluator-for Mistral AI | official |
| VALS AI | reach | 6 | output acted on — Anthropic's Sonnet 4.5 launch footnote: 'Finance Agent: All scores reported were run and published by Vals AI on their public leaderboard'; its May 2026 finance post says Opus 4.7 'leads the industry on Vals AI's Finance Agent benchmark'. | first-hand 2025-09-29 |
| VALS AI | reach | 6 | output acted on — Vals AI contract Center for AI Standards and Innovation (NIST) | official |
| VICTORIA KRAKOVNA | durability | 6 | built operating — FLI's own About page says the institute was initially run by volunteers coordinated by co-founder Victoria Krakovna and now has 20+ staff. | first-hand |
| VICTORIA KRAKOVNA | durability | 4 | cited work — Her master list of specification gaming examples, published 2 April 2018 and kept current, is the reference list others collect examples into. | first-hand 2018-04-02 |
| VICTORIA KRAKOVNA | insulation | 2 | at will employee — Victoria Krakovna employed-by DeepMind | official |
| VICTORIA KRAKOVNA | leverage | 4 | advisory seat — Victoria Krakovna board-seat Future of Life Institute | official |
| VIRALITY PROJECT | entrenchment | 2 (ended) | reports persist — The Virality Project's final report, Memes, Magnets, and Microchips, and its public weekly briefings remain available after the project's newest briefing of August 2021. | first-hand 2022-02-24 |
| VIRALITY PROJECT | reach | 6 (ended) | output acted on — The final report states 'the Office of the Surgeon General incorporated VP's research and perspectives into its own vaccine misinformation strategy' (the project's own account). | first-hand 2022-02-24 |
| WHO + GAVI DIGITAL-HEALTH STACK | entrenchment | 8 | program of record — WHO's Global Digital Health Certification Network, with WHO as trust anchor, is described on WHO's own event page as already connecting over 80 countries and nearly 2 billion people. | first-hand |
| WHO + GAVI DIGITAL-HEALTH STACK | reach | 10 | statutory power — WHO states the International Health Regulations are an instrument of international law legally binding on 197 countries, current since 19 September 2025 with the amendments the World Health Assembly adopted in 2024 (resolution WHA77.17). | first-hand 2025-09-19 |
| WORLD ECONOMIC FORUM | entrenchment | 6 | operating — The Forum has operated since 1971 as a Swiss foundation and has held Swiss Host-State Act status as an International Institution since 2015. | first-hand 2015-01-23 |
| WORLD ECONOMIC FORUM | funding insulation | 4 | graded funding disclosed — Consolidated revenue for the year to 30 June 2026 was CHF 461,305 thousand, of which partnership and membership income was CHF 297,671 thousand (about 65 percent) and participation fees CHF 41,168 thousand, all from the partners and attendees it serves. | first-hand 2026-06-30 |
| WORLD ECONOMIC FORUM | reach | 4 | convenes or funds — The Forum describes itself as the international organization for public-private cooperation and convenes the Davos Annual Meeting and 973 partners worldwide at 30 June 2026. | first-hand 2026-06-30 |
| WORLDCOIN / TOOLS FOR HUMANITY / WORLD ID | entrenchment | 8 | program of record — World states 'The real human network now spans nearly 18 million verified humans' (company-reported, not audited). | first-hand 2026-04-01 |
| WORLDCOIN / TOOLS FOR HUMANITY / WORLD ID | entrenchment | 6 | operating — World's about page states Tools for Humanity is privately held, with Sam Altman as chairman and Alex Blania as CEO, and it operates the Orb network. | first-hand |
| WORLDCOIN / TOOLS FOR HUMANITY / WORLD ID | reach | 6 | output acted on — World's own page says a Human Badge means a verified World ID has been connected to a Tinder account (World ID for Tinder). | first-hand |
| XAI (SPACEXAI) | entrenchment | 8 | program of record — The S-1 states Anthropic agreed in May 2026 to pay $1.25 billion per month through May 2029 for compute capacity across COLOSSUS and COLOSSUS II (terminable on 90 days' notice). | first-hand 2026-05-01 |
| XAI (SPACEXAI) | entrenchment | 6 | operating — xAI operates the Grok models and the xAI Gov offering as a subsidiary of SpaceX, per the S-1 and its own site. | first-hand |
| XAI (SPACEXAI) | funding insulation | 10 | controlled or statutory — SpaceX's S-1 states Mr. Elon Musk had 'a controlling financial interest in the Company, xAI and X through his majority voting interest in each such entity', and that xAI is now a wholly-owned subsidiary of the Company. | first-hand |
| XAI (SPACEXAI) | reach | 8 | contract or gate — xAI states it was selected by the U.S. Department of War to deliver frontier AI for military and national security users. | first-hand 2025-12-22 |
| YANN LECUN | durability | 6 | built operating — AMI Labs' launch post of 10 March 2026 says 'We've raised a $1.03B USD (~EUR 890M) round from global investors', co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital and Bezos Expeditions. | first-hand 2026-03-10 |
| YANN LECUN | durability | 6 | built operating — Yann LeCun co-founder Partnership on AI | official |
| YANN LECUN | insulation | 3 | appointed officer — TechCrunch describes AMI Labs as 'co-founded by Turing Award winner Yann LeCun', refers to 'LeCun's involvement as chairman', and names Alexandre LeBrun as CEO. | press 2026-03-09 |
| YANN LECUN | leverage | 4 | advisory seat — Yann LeCun board-seat AI Alliance | official |
| YANN LECUN | leverage | 2 | public voice — The Senate Select Committee on Intelligence hosts LeCun's written testimony of 19 September 2023 (page returned 403 to fetch; search results describe it as identifying him as co-founder of Meta's FAIR and Meta's Chief AI Scientist). | official 2023-09-19 |
| YOEL ROTH | durability | 2 | publications — His sworn House Oversight statement of February 2023 remains in the public record. | official 2023-02-08 |
| YOEL ROTH | durability | 1 | publications — Yoel Roth co-author Renee DiResta | scholarly |
| YOEL ROTH | insulation | 4 | appointed officer — Match Group's team page lists him as Head of Trust & Safety (Vice President); no term or protection is stated. | first-hand |
| YOEL ROTH | leverage | 8 | signs enforcement — Match Group's team page lists him as Head of Trust & Safety for Tinder, Hinge and other dating apps. | first-hand |
| YOEL ROTH | leverage | 8 (ended) | signs enforcement — His House Oversight statement says he was Head of Trust & Safety at Twitter and that his team enforced its written rules. | official 2023-02-08 |
| YOEL ROTH | leverage | 2 | public voice — Yoel Roth testified U.S. House of Representatives | official |
| YOSHUA BENGIO | durability | 6 | built operating — Yoshua Bengio founded LawZero | official |
| YOSHUA BENGIO | durability | 6 | built operating — Mila's own release calls him Founder and Scientific Director of Mila and Chair of the International Scientific Report on the Safety of Advanced AI. | first-hand 2025-01-29 |
| YOSHUA BENGIO | durability | 4 | cited work — The International AI Safety Report, chaired by him, is produced with a UK Government secretariat for governments' use. | first-hand 2025-01-29 |
| YOSHUA BENGIO | leverage | 6 | runs org acted on — The International AI Safety Report site says it was written by over 100 independent experts and led by Yoshua Bengio, as a shared picture for decision-makers in government. | official |
| YOSHUA BENGIO | leverage | 4 | advisory seat — Yoshua Bengio advisory METR (Model Evaluation & Threat Research) | official |
| YOSHUA BENGIO | leverage | 4 | advisory seat — Yoshua Bengio advisory Transluce | official |
| YOSHUA BENGIO | leverage | 4 | advisory seat — The Prime Minister of Canada named him a member of the new national council on artificial intelligence on 2 October 2026. | official 2026-10-02 |
| J. ZICO KOLTER | durability | 6 | built operating — Gray Swan's about page lists Zico Kolter as co-founder and Chief Scientist of a firm still operating. | first-hand |
| J. ZICO KOLTER | insulation | 4 | appointed officer — The Delaware AG statement makes the SSC Chair a director on the nonprofit board and not on the PBC board; the nonprofit appoints and can remove PBC directors, and nothing protects his seat. | official 2025-10-28 |
| J. ZICO KOLTER | leverage | 8 | signs deployment — The Delaware AG's 28 October 2025 statement says the SSC, chaired by Zico Kolter, has the authority to require mitigation up to and including halting the release of models or AI systems. | official 2025-10-28 |
| J. ZICO KOLTER | leverage | 6 | runs org acted on — Gray Swan's Series A release lists him as co-founder and Chief Scientist and says the firm is cited in 11 recent frontier model system cards including Anthropic, OpenAI and Meta. | first-hand 2026-05-28 |
| J. ZICO KOLTER | leverage | 4 | advisory seat — Zico Kolter advisory UK AI Security Institute | official |
| J. ZICO KOLTER | leverage | 4 | advisory seat — Zico Kolter board-seat OpenAI Foundation | official |