GILLIAN HADFIELD

Behavioral Archetype

THE LEGAL ECONOMIST WHO DESIGNED THE LICENSE — Subject is the academic who proposed that governments license private regulators and require the regulated to buy their services, then spent seven years moving that design from a 2017 book, to a 2019 paper with Jack Clark, to a nonprofit that sponsored the California statute creating state-designated independent verification organizations, to a federal bill, to the Governor’s expert group. Her own home page says the research “has helped shape the Independent Verification Organization (IVO) model, a concrete instantiation of that approach that has been introduced in state and federal legislation” (Gillian K. Hadfield). Fathom’s release of 23 September 2026 calls her “key architect of the IVO (Independent Verification Organization) framework” (Fathom, 23 September 2026). This profile scores the position, not the person.

The position is the seam between the author of a market design and the organizations that sell into it. She is a professor with no seat that licenses anyone. She is also the senior advisor of the Fathom that sponsored the statute, an advisor to AVERI, which Fathom funds, a keynote speaker for FAR.AI, and a member of the group advising the Governor on the executive order that moves the designation criteria forward. Every one of those ties is on a page she or the other party published. None is evidence of what any designation committee will do, and nobody has been designated.

Essence Indicators

• Bloomberg Distinguished Professor of AI Alignment and Governance, Johns Hopkins University. Her biography: “Bloomberg Distinguished Professor of AI Alignment and Governance,” at the School of Government and Policy and the Whiting School of Engineering (Gillian K. Hadfield, bio). The Governor’s office gives the same affiliation: “professor at Johns Hopkins University with a focus on AI alignment and innovative design for legal, regulatory, and technical systems for AI” (Governor, 23 September 2026). Her CV lists the post under “Present Positions” and under honors and awards (CV).
• Senior Advisor, Fathom. Her CV: “FATHOM AI INC. Senior Advisor, 2025 – present” (CV). Fathom’s release: “key architect of the IVO (Independent Verification Organization) framework and Fathom senior advisor, Dr. Gillian Hadfield,” with CEO Andrew Freedman saying “Two years ago we teamed up, and Fathom has proudly worked alongside Dr. Hadfield to bring that idea to life” (Fathom, 23 September 2026). Fathom’s team page lists her under “Affiliate Members” as “Bloomberg Distinguished Professor of AI Alignment and Governance” (Fathom, About). Her bio: she “promotes better AI policy with the non-profit Fathom” (bio).

• Author of the regulatory-markets proposal. Her bio: the 2017 book Rules for a Flat World “introduced” regulatory markets, “dynamic, democratically-accountable systems that enable regulation to keep pace with rapid technological innovation,” and “In 2026 she published Regulatory Markets: The Future of AI Governance in Jurimetrics” (bio). The arXiv record lists the paper by Hadfield and Jack Clark, submitted 11 April 2023 and revised 3 February 2026, with the journal reference “Jurimetrics: The Journal of Law, Science and Technology, Volume 65 pp. 195-240 (2026)” (arXiv:2304.04914). The earlier paper by Clark and Hadfield, “Regulatory Markets for AI Safety,” was submitted 11 December 2019: “We propose a new model for regulation to achieve AI safety: global regulatory markets” (arXiv:2001.00078). A 2026 PNAS Perspective: “I review three examples I have proposed: the creation of registration regimes for frontier models; the creation of registration and identification regimes for autonomous agents; and the design of regulatory markets” (arXiv:2602.01474).

• The Governor’s group. On 23 September 2026 the Governor announced “the group of leading experts who will advise on recommendations required by his recent AI executive order,” naming four: Jason Goldman, Gillian Hadfield, Alondra Nelson and Rob Reich. “Proposals under consideration include requiring independent third parties to be embedded in frontier AI companies to verify safety frameworks and independently assess safety evaluations” (Governor, 23 September 2026). Her statement in the release: “The need for effective, accountable, and independent oversight of increasingly powerful and autonomous AI is urgent and I’m very glad to join this great team to quickly help ensure we steer in the right direction.”
• Canada’s National Council on Artificial Intelligence. The Prime Minister’s release of 2 October 2026 lists her among the members of a council that “will bring together public and private stakeholders and advise on the evolution of the government’s AI for All strategy,” beside Yoshua Bengio (Prime Minister of Canada; Mark Carney).
• Former Senior Policy Advisor, OpenAI. Her bio: “From 2018 to 2023 she was a Senior Policy Advisor for OpenAI.” Her CV lists “OPENAI Senior Policy Advisor, 2018 – 2023” among other positions held, and a visiting-researcher position now: “Visiting Faculty Researcher (part-time) in the Paradigms of Intelligence (Pi) Group at Google” (bio; CV).
• Inaugural director, Schwartz Reisman Institute. She “previously served as the inaugural director and held the Schwartz Reisman Chair in Technology and Society” at the University of Toronto; the CV gives the institute directorship as 2019 to 2023 and the chair as 2019 to 2024, with the chair funded at “$500,000” (bio; CV). The institute’s founding donors, Gerald Schwartz and Heather Reisman, “made a generous gift in 2019 that made it possible for the University to establish” it (Schwartz Reisman Institute, founding donors). She remains “professor of law (status-only)” at Toronto.
• Advisor, AVERI. AVERI’s team page lists her among its Advisors: “Gillian K. Hadfield is an economist and legal scholar turned AI researcher” (AVERI, Team). See Bri Treece, Fathom’s President, also an AVERI advisor.
• Other seats. “Chair, Board of Trustees, Cooperative AI Foundation”; “Board of Advisors, Americans for Responsible Innovation”; “Board of Advisors, International Association for Safe and Ethical AI”; “Advisor, International Dialogues on AI Safety”; “Member, RAND Technology Advisory Group”; and a Schmidt Sciences AI2050 Senior Fellowship (bio; CV). See RAND and Schmidt Sciences.

On the Record

Her words, from her own site, her papers, a June 2026 op-ed and a workshop.

On the problem, 2023. “First is a technical deficit: Legislatures and regulatory face significant challenges in rapidly translating conventional command-and-control legal requirements into technical requirements. Second is a democratic deficit: Over-reliance on industry to provide technical standards fails to ensure that the many values-based decisions that must be made to shape AI development and deployment are made by democratically accountable public, not private, actors” (arXiv:2304.04914).

On the design. “Governments license and audit private regulators, evaluating their regulatory services against the required outcomes.” The regulated “are required by governments to purchase the regulatory services of appropriate licensed private regulators” (Hadfield and Clark, Jurimetrics).

On its limits, in the same paper. “We do not think that regulatory markets will work in all circumstances.” “Regulatory capture is a known risk in existing government-led regulation.” “Regulatory markets put an additional layer between governments and industry. This creates a risk that private regulators, which are trying to sell their services to AI companies, will collaborate with those companies to cheat on government goals.” The protection she names: “effective threats to condition, suspend, or revoke the licenses of regulators that skimp on performance to win the business of targets.” And: “But regulatory markets only work if governments are willing to regulate private regulators” (Hadfield and Clark, Jurimetrics).

On the cautionary cases, same paper. Of credit rating agencies in 2008: “these regulatory intermediaries were deliberately shielded from liability for errors in their ratings and there was no formal oversight from government.” And on thin markets: “If there are only two or three companies involved in developing a particular type of AI, it will be difficult to sustain a competitive market of regulators” (Hadfield and Clark, Jurimetrics).

On the sector it would build, 19 June 2026. In the Washington Examiner, on the Obernolte-Trahan bill: “It builds on a model for regulating complex, fast-moving technologies that my colleagues at Fathom and I developed a decade ago: the Independent Verification Organization, or IVO.” “Why license private verifiers — why not just fund CAISI to run the tests? Because the tests don’t exist yet.” “The strongest independent work comes from a few small outfits such as METR, Apollo Research, and Transluce that survive on whatever access they can negotiate. The IVO model turns it into a genuine sector, pulling investment and talent toward the measurement science we lack.” “Tens of billions flow into building artificial intelligence; a rounding error goes into overseeing it.” The byline reads: “Gillian Hadfield is the Bloomberg Distinguished Professor of AI Alignment and Governance at Johns Hopkins University” (Washington Examiner). See METR, Apollo Research and Transluce.

On markets, April 2024 workshop. “Ideological arguments aside, markets are good for solving problems, and when properly regulated, they are effective at attracting resources where they’re needed” (Schwartz Reisman Institute).

Social Persona / Impression Management

Immediate impression: The institutional designer. Her home page presents “Building AI That Understands Human Systems” and describes the work as “developing the legal, regulatory, and technical infrastructure for artificial intelligence that operates within human normative frameworks” (Gillian K. Hadfield). The pages carry papers, a book, chairs, and a CV of grants, talks and committees.

Energy: Academic with a policy-practitioner’s output. Her CV lists keynotes at the FAR AI Alignment Workshop in London (March 2026), a panel for the Ohio House’s technology committee, and a meeting with Government of Canada officials “re: Regulatory Markets” (CV). Her site links X and Bluesky accounts and a press contact at Johns Hopkins.

Impression management strategy: The idea is introduced as research and then as a movement from theory to practice. Her policy page: “This approach is now moving from theory to practice,” and “I am working with Fathom to develop Independent Verification Organizations (IVOs)—government-licensed private entities that verify AI systems against specific risk claims through ongoing monitoring—as the central mechanism for building regulatory markets for AI” (Gillian K. Hadfield, policy). Her June 2026 byline names her Johns Hopkins chair and not her Fathom role, while the text of the piece names Fathom (Washington Examiner). Her bio frames the book as the origin and the statute as a descendant: “elements of which have since been incorporated into the proposed Great American AI Act” (bio).

Forensic Archetype Comparison

Psychometric Assessment

Trait scores are observations of the documented public role, not claims about private character. The public record is a CV, two sites, a book’s worth of papers, one op-ed and the appointing bodies’ pages.

Big Five (OCEAN) (inferred from the public role only):

Dark Triad (evidence-bound; the score measures structural position, not character):

MBTI: ENTP-adjacent, inferred from the documented public role only. The rationale is the work product: an institutional inventor who keeps moving one design across disciplines and jurisdictions. Her policy page states the method: “I develop practical proposals for governing AI” (Gillian K. Hadfield, policy). A style inference from her writing, not a test result.

Threat Assessment

Score move: none. Calibrated against Bri Treece (5/5/6, hazard 53.3), President of the organization that sponsored the statute and a board member of an evaluator, who holds one point less leverage because she authored no design and two less insulation because her income is a Fathom salary; against Percy Liang (7/8/8, hazard 70), whose benchmarks governments run and whose center has many named funders, a point above on each axis; against Tino Cuellar (7/8/8, hazard 74), with whom she lists a collaboration on legal infrastructure and who holds a trustee seat at a lab’s governance trust, a point above on each axis; and against Helen Toner (6/8/8, hazard 80), level on leverage, one point more insulated on a philanthropic general-support base, and one point more durable on a centre that outlasts its director.

Score basis. Atk 6 is leverage: she is credited by the sponsor as “key architect” of the IVO framework and sits on the Governor’s group; the points under a higher number are that she holds no designation, licensing or release power, the chapter does not “Require any person, partnership, or corporation that develops, deploys, or operates an AI system or model to engage an IVO,” and the federal bill is introduced and not law (Fathom, 23 September 2026; leginfo, SB 813). Def 7 is insulation: a named chair and grants listed from Schmidt Sciences ("$999,900, 2023–2025"), the John Templeton Foundation ("$2.2 million, 2026–2029"), Meta ("$675,000, 2022–2023") and others; the points off are a part-time Google appointment, a Meta grant, an undisclosed Fathom arrangement and no stated majority funder either way (CV). Hp 7 is durability: a 2017 book, a 2019 paper, a 2026 Jurimetrics edition and a chaptered statute whose text defines the class of designated verifiers; the points off are that the statute is voluntary, its criteria are due by 1 January 2028 and the federal bill is not law (leginfo, SB 813; arXiv:2304.04914).

Alignment Analysis

Stated alignment: Governments should license private regulators, and the regulated should buy their services, so that regulation keeps pace with the technology while “values-based decisions” stay with “democratically accountable public, not private, actors” (arXiv:2304.04914; Hadfield and Clark, Jurimetrics). Observed alignment: She has put the design into a book, two papers and a PNAS Perspective, advises the organization that sponsored the California statute, keynotes at an evaluator’s workshop on it, briefs Ohio legislators and Canadian officials, and sits on the Governor’s group under the order that accelerates it (CV; Governor, 23 September 2026). Gap assessment: None on direction. The statute departs from the paper at points the paper’s abstract and its risk list name, and she has said in print where the design breaks. The record raises a structural question and does not settle it: the author of the market design advises the sponsor of the statute, an auditor the sponsor funds and the Governor’s group, each seat disclosed. The pages show disclosed seats and nothing more, and for a file that scores a position it never needs to.

The design, tie by tie

Her paper’s elements set beside the chaptered statute (Chapter 179, Statutes of 2026) and the federal bill:

The statute adopts the license, the independence criteria and the revocation procedure, and drops the mandatory purchase. The federal bill, on her own description, restores the purchase for the largest developers and, on the text, adds an immunity clause that her paper’s rating-agency warning cuts against.

Interest: Documented, never motive, never a verdict. She argues three positions.

1. Position: government-licensed private verifiers, as the central mechanism of AI regulation (“I am working with Fathom to develop Independent Verification Organizations (IVOs)… as the central mechanism for building regulatory markets for AI,” Gillian K. Hadfield, policy). What adoption pays: the statute creates a designated class, the IVO, an “AI auditor” being “a person, partnership, academic institution, nonprofit, or corporation that conducts a covered audit on behalf of a third party” (leginfo, SB 813); a nonprofit evaluator that conducted such audits would fall within it, among them the outfits her op-ed names as surviving “on whatever access they can negotiate,” METR, Apollo Research and Transluce, and AVERI, which she advises. She is Senior Advisor to the sponsor; her CV lists a Fathom workshop at IASEAI in Paris and a keynote on IVOs at the FAR AI Alignment Workshop in London in March 2026; no designation, fee, contract or equity paid to her is documented, and the CV states no compensation for the Fathom role (CV; Fathom, 23 September 2026). Her best case: “Why license private verifiers — why not just fund CAISI to run the tests? Because the tests don’t exist yet,” and “Tens of billions flow into building artificial intelligence; a rounding error goes into overseeing it” (Washington Examiner). Others who gain: the evaluators; legislators who need not write technical tests into law; and the labs, among them OpenAI, which endorses “SB 813 on overall infrastructure for independent safety assessments” (OpenAI) and where she was a Senior Policy Advisor to 2023, and Anthropic, whose co-founder is her co-author.
2. Position: verification should be mandatory for the largest developers (“The Obernolte-Trahan bill makes verification mandatory for the most powerful models and puts the government in charge of the standard,” Washington Examiner). What adoption pays: a statutory buyer: H.R. 9925 provides that “a very large frontier developer shall retain an independent verification organization that has been licensed,” which turns a voluntary market into a purchase obligation for the largest developers (Congress.gov, H.R. 9925). Her best case: “That’s progress, but it leaves them grading their own homework” of transparency laws in California, New York and Illinois (Washington Examiner). Others who gain: every licensed verifier; and the federal standards body, CAISI, which her op-ed names as setting the acceptable-risk standard.
3. Position: legal infrastructure, meaning registration of frontier models and identification of agents, before substantive rules (“a key role for law is not only to establish substantive rules but also to establish legal and regulatory infrastructure to generate and implement rules,” arXiv:2602.01474). What adoption pays: registration regimes create the record against which verification is run; her policy page lists the work with Tino Cuéllar and Tim O’Reilly (Gillian K. Hadfield, policy). No fee or contract is documented. Her best case: her site describes the paper as arguing “that effective AI governance depends less on substantive rules than on legal infrastructure” (Gillian K. Hadfield). Others who gain: any registry operator and every agency that gains visibility into model development.

Documented critiques, in their authors’ words. The record contains no academic rebuttal of her paper that this file found; the written critiques are of the designation design, in committee analyses and in the paper’s own risk list. The coalition of the California Initiative for Technology & Democracy, the Children’s Advocacy Institute and Tech Equity on the 2025 version: “It is not clear whether there are any entities currently in the United States other than perhaps the sponsor of this legislation who might be able to comply with the current MRO qualifications in the measure” (Senate Judiciary analysis, 29 April 2025). TechNet, same analysis: “Rather than establishing bespoke organizations and frameworks from scratch, we encourage California to build on the work of established, globally respected standards-setting bodies.” TechNet on the June 2026 version: “a state-specific definition of “substantially exceeds compliance” adds to a growing patchwork rather than reducing it”; the California Chamber of Commerce: “No seven individuals can reasonably be expected to possess the collective expertise necessary to evaluate technical standards, auditing methodologies, and risk-assessment frameworks across all of these domains simultaneously” (Assembly Privacy analysis, 1 July 2026); the June text with its commission was not enacted. On the independence premise, Fathom’s own July 2026 piece quotes Rumman Chowdhury: “Being a purely independent evaluator is nearly impossible, because it’s just not a fully built-out ecosystem yet” (Fathom, Who Evaluates the Evaluators?). Her own paper names the same failure modes: “Regulatory capture is a known risk,” the thin-market problem, and that “regulatory markets only work if governments are willing to regulate private regulators” (Hadfield and Clark, Jurimetrics).

Alliance Map

The overlap is the finding. At one time the same person is the credited author of the verifier-market design, the senior advisor of the organization that sponsored the statute adopting it, an advisor to the auditor Fathom funds, a keynote speaker at an evaluator’s workshop on it, a member of the group advising the Governor on the order that moves its criteria forward, a member of Canada’s national council, a former advisor to a lab whose policy supports the statute, and a co-author with a lab’s co-founder. Funders listed on her CV, among them Schmidt Sciences, are also on the evaluator map. The documents show disclosed seats, listings and grants, not coordination, and no designation has been made.

Convergent Drive Classification

Self-preservation: Holds concurrent chairs and seats across a university, a lab-adjacent research role, a nonprofit advisory role, two governments’ advisory bodies and a foundation board, so no single loss removes the program (CV). Goal preservation: Keeps one design moving from book to paper to statute to federal bill to executive order, and revises the 2023 preprint in 2026 as the legislation moves (arXiv:2304.04914). Resource acquisition: Draws grants from named philanthropic, corporate and public sources, among them Schmidt Sciences, the John Templeton Foundation, Meta, the Laude Institute and Johns Hopkins (CV). Self-improvement: From a legal economist’s book chapter to an advisory seat on the Governor’s group and a seat on a national council (Governor, 23 September 2026; Prime Minister of Canada).

Subject is not an AI system. The drives appear anyway — in the architect of the license.

Public footprint: gillianhadfield.org · X, @ghadfield · Bluesky.

Sources: Gillian K. Hadfield — home; Gillian K. Hadfield — bio; Gillian K. Hadfield — policy; Gillian K. Hadfield — research; Hadfield CV, August 2026; Hadfield and Clark — Regulatory Markets: The Future of AI Governance (arXiv:2304.04914); Hadfield and Clark — Jurimetrics text (PDF); Clark and Hadfield — Regulatory Markets for AI Safety (arXiv:2001.00078); Hadfield — Legal Infrastructure for Transformative AI Governance (arXiv:2602.01474); Washington Examiner — The uncomfortable truth behind AI debate; Fathom — Hadfield appointed, 23 September 2026; Fathom — About; Fathom — Who Evaluates the Evaluators?; AVERI — Team; Governor — experts announced, 23 September 2026; Prime Minister of Canada — National Council on Artificial Intelligence, 2 October 2026; Schwartz Reisman Institute — co-designing regulatory markets; Schwartz Reisman Institute — founding donors; leginfo — SB 813 text; Congress.gov — H.R. 9925 text; Senate Judiciary Committee analysis of SB 813, 29 April 2025; Assembly Privacy and Consumer Protection analysis of SB 813, 1 July 2026; OpenAI — AI policy window.