JASON MATHENY

Behavioral Archetype

THE THREAT MODELER AT THE FOUNDRY — Subject spent a decade inside the intelligence community’s research arm, founded the Georgetown center that supplied the Biden administration’s technology staff, served as the White House’s dual-hatted coordinator for technology and national security, and then took the presidency of the think tank that holds four federal research contracts whose researchers, on POLITICO’s 2023 reporting, helped write the AI executive order. In the three years since he has told four congressional committees what rules to write for compute, model release and synthetic DNA, and he directs philanthropic gifts to RAND’s AI and biosecurity work. This profile scores the seat, a research chief who sits at the point where a threat assessment becomes a rule and where the money for the assessment arrives, and not the man.

Documented conduct. The pattern is the analyst’s: describe the danger in the language of national security, then attach a control. His 8 March 2023 testimony names “the development of novel cyber weapons, large-scale disinformation attacks, and the design of advanced biological weapons” and proposes eight actions, among them “consider creating a regulatory framework for AI” and tracking “where high-performance computing hardware goes and what it is being used for” (S. Hrg., Senate Homeland Security, 8 Mar 2023). On 19 April he went further: “I think we need a licensing regime, a governance system of guardrails around the models that are being built” (Senate Armed Services Cybersecurity Subcommittee, 19 Apr 2023). On 22 June, asked by Rep. Babin how to square open research with security, he said “before publishing or open sourcing the largest models, I recommend that we require some amount of red teaming and evaluation” (House Science Committee, 22 Jun 2023).

Essence Indicators

• President and Chief Executive Officer of the RAND Corporation. RAND’s press release of 7 June 2022 says he “has been named president and chief executive officer,” effective 5 July, “the sixth president and CEO,” succeeding Michael D. Rich, and that the board “considered nearly 200 potential leaders” (RAND, 7 Jun 2022). RAND’s leadership page lists him first among the organizational leadership (RAND, Leadership).

• RAND’s own bio: before RAND “he led White House policy on technology and national security at the National Security Council and the Office of Science and Technology Policy”; earlier “founding director of the Center for Security and Emerging Technology at Georgetown University and director of the Intelligence Advanced Research Projects Activity (IARPA)”; before that “Oxford University, the World Bank, the Applied Physics Laboratory, the Center for Biosecurity, and Princeton University.” Degrees: Ph.D. in applied economics and M.P.H., Johns Hopkins; M.B.A., Duke; B.A. in art history, Chicago (RAND, Jason Matheny).
• The White House titles are on the day-of record: “deputy assistant to the president for technology and national security, deputy director for national security in the Office of Science and Technology Policy and coordinator for technology and national security at the National Security Council,” per FedScoop of 9 March 2021 (FedScoop). In his own account of the job: “The role was created to help bridge the National Security Council and the Office of Science and Technology Policy” (RAND, Q&A with Jason Matheny, 8 Mar 2024).

• IARPA: Nextgov reported on 4 August 2015 that “Matheny is taking over as IARPA’s director,” having “come to the Intelligence Advanced Research Projects Activity as a program manager in 2009” and led its “Anticipating Surprise” office (Nextgov, 4 Aug 2015). His own House Intelligence statement lists “Assistant Director of National Intelligence” and a seat as “a Commissioner on the National Security Commission on Artificial Intelligence, to which I was appointed by the Senate Select Committee on Intelligence” (HPSCI statement, 28 Feb 2023).
• Earlier career, from the alumni magazine: “in 2004 he founded New Harvest, the world’s first nonprofit organization dedicated to supporting in vitro meat research,” run from “his Baltimore apartment” while he was a Hopkins doctoral student (University of Chicago Magazine, Sept 2006).
• Trustee of Anthropic’s Long-Term Benefit Trust, September to December 2023. The Trust’s founding post lists “Jason Matheny: CEO of the RAND Corporation” among its first trustees and records that “In December 2023, Jason Matheny stepped down from the Trust to preempt any potential conflicts of interest” (Anthropic, The Long-Term Benefit Trust). The post says the Class T stock gives the Trust “the authority to elect and remove a number of Anthropic’s board members” (same). The seat lasted about three months.

• Gifts at his discretion. Open Philanthropy’s grant pages: “Open Philanthropy gave a gift of $2.5 million to RAND Corporation, to be spent at the discretion of RAND president and CEO Jason Matheny,” allocated across “a technology policy training program,” “support for the Pardee RAND Graduate School,” “a new research center focused on China studies” and “a research fund that will help to produce information for policymakers about emerging technology and security priorities,” recommended by Luke Muehlhauser (24 July 2025). A second, “a gift of $3.4 million over two years,” went “toward the development of a forecasting platform that will be used by professional forecasters and RAND staff to inform RAND’s work” (17 July 2025, recommended by Benjamin Tereick) (Coefficient Giving grants feed). The $10.5 million of October 2023 and the $5.5 million and $10 million POLITICO reported for spring 2023 are on the RAND profile; Sentinel Bio lists $445,000 (November 2025) and $756,000 (May 2026), each “to be spent at the discretion of RAND president and CEO Jason Matheny.”

• The executive order. POLITICO reported on 16 December 2023 that RAND was “a driving force” behind the White House’s AI reporting requirements and “played a key role in drafting” Executive Order 14110 (POLITICO, Brendan Bordelon, 16 Dec 2023). In September 2024 Sen. Ted Cruz, then the ranking member of the Senate Commerce Committee, wrote to him about RAND’s role in drafting the order (Senate Commerce Committee). RAND’s answer, as Fox News relayed it in September 2025: Matheny had “no role” in crafting the order, and its “researchers did provide technical expertise and analysis to inform the EO in response to requests from policymakers” (Fox News, 8 Sep 2025).

• Effective altruism, as reported. Fox News calls him “a leader in the effective altruism movement” and quotes a former Defense Department official saying he “been very deliberate about inserting personnel who share his AI-doomerism worldview”; RAND’s spokesperson answered that AI as an “existential threat” is “not the lens” RAND uses, and that he “remains committed, along with RAND, to contributing expertise and analysis” to the Trump administration (Fox News, 8 Sep 2025). The documented ties are in the Alliance Map under Movement.

• On the record about extinction risk and open release. In a June 2023 interview published by ChinaTalk on 8 December 2023 he said “I saw the greatest degree of unanimity that I’ve seen among researchers describing AI as an extinction risk” and, of release choices, “thinking carefully about open-sourcing, because that’s something you can’t really take back” (ChinaTalk, Jordan Schneider, 8 Dec 2023).
• Federal giving. The FEC’s itemized-receipts file returns no individual contribution under “Matheny” with RAND or the Center for Security and Emerging Technology as employer (FEC API, queried 24 October 2026).

Social Persona / Impression Management

Immediate impression: A technocrat with the bearing of a civil servant: economist, intelligence-community program manager, White House staffer. RAND’s board chair describes him in the hiring release as “a compelling speaker and a serious listener” and “a generous and inclusive leader” (RAND, 7 Jun 2022). In print he is plain and numbered: eight actions, five actions, six actions, each a recommendation a staffer can lift into a bill.

Energy: Measured and forward-looking. The 2023 statements all open the same way, “Among a broad set of technologies, AI stands out both for its rate of progress and for its scope of applications,” and then move to a list of controls (House Science Committee, 22 Jun 2023). The register shifts only when the subject is China, where he talks about “red teams” modeled on Soviet studies (RAND, Q&A, 8 Mar 2024).

Impression management strategy: The analyst’s method: describe the threat as unprepared-for (“grave security challenges for which we are currently unprepared”), name a control, and attach a commercial frame so the control reads as good for the industry it binds. To the House Science Committee he said “many AI firms are calling for government oversight to ensure that AI systems are safe and secure: It’s good for their business” and “The United States can make safety a differentiator for our AI industry” (House Science Committee, 22 Jun 2023). The 16 August 2023 essay he wrote for the Washington Post, republished by RAND, uses the March 2023 LLaMA leak as the proof case: “Meta has been unable to remove LLaMA from the internet” (RAND, A Model for Regulating AI).

• Social accounts and feeds. X @JasonGMatheny and a LinkedIn profile, both linked from his RAND expert page (RAND, Jason Matheny); RAND’s page lists his commentary and testimony by date. No personal blog, newsletter or RSS feed was located on his RAND page; the RAND expert page is his own publication list.

Forensic Archetype Comparison

Psychometric Assessment

Scores are observations of the documented public role, drawn from his own testimony and writing, and are not claims about private character.

MBTI: INTJ (the planner who writes the controls; a read of the role, not a diagnosis).

Threat Assessment

The hazard is the position, not the man: the same person’s sworn testimony, his institution’s funding and the state’s draft language sit in one chain, and the record can establish each link without establishing any influence along it. Score basis: atk 7 and hp 7 sit one under RAND (8 and 9); def 5 is below it because a single executive’s insulation is lower than an institution’s.

Alignment Analysis

Stated alignment: Make AI and synthetic biology safe for national security and competitiveness through controls on hardware, training and release: “The United States can make safety a differentiator for our AI industry, just as it was a differentiator for our early aviation and pharmaceutical industries” (House Science Committee, 22 Jun 2023); “Systems that exhibit dangerous capabilities should not be released until safety can be assured” (RAND, A Model for Regulating AI).

Observed alignment: The recommendations recur across four forums and each maps to work RAND does or funders have paid for: intelligence-community strategies and “National Intelligence Estimates and Net Assessments” on synthetic biology and AI (HPSCI statement, 28 Feb 2023); defense contracts with know-your-customer and security terms (S. Hrg., 19 Apr 2023); a forecasting platform and a China center funded at his discretion (Coefficient Giving); nucleic-acid screening work funded by Sentinel Bio (Sentinel Bio).

Gap assessment: Stated and observed alignment coincide on the record and diverge on who benefits. The record cannot settle whether the recommendations follow from the analysis or from the institution’s position, and for this seat it does not need to: the testimony and the funding are each on the page, labelled, and the reader can price one against the other.

Interest: Documented, never motive, never a verdict.

• Position: licensing and know-your-customer rules for the AI supply chain (“I think we need a licensing regime”; know-your-customer screening “for all customers before training large AI models,” in federal cloud and developer contracts). What adoption pays: the rule needs “rigorous review by a regulator or third-party evaluator” before a model is released (RAND, A Model for Regulating AI); RAND runs four federal research centers and an AI, security and technology center that the same funders back, so a review regime is demand for the kind of analysis and red-team work it sells to the state. His best case: “if an AI is stolen or leaked even once, it could be impossible to prevent it from spreading throughout the world” (same). Who else gains from the same rule: the largest developers, whom he says are asking for it (“many AI firms are calling for government oversight… It’s good for their business”), and any lab that can afford compliance; Anthropic asks Congress to codify government pre-deployment testing, through its Frontier Red Team head (Logan Graham).
• Position: caution on open-weight release (“especially cautious about open sourcing large language models that could be misused”; “before publishing or open sourcing the largest models… require some amount of red teaming and evaluation”). What adoption pays: a pre-release evaluation gate falls hardest on developers who publish weights, and the House Science panel he spoke on included the chief executive of Hugging Face, the open-model hub, and the gate applies first to whoever publishes weights (House Science Committee, 22 Jun 2023). His best case: open-sourcing “is something you can’t really take back” (ChinaTalk), with the LLaMA leak as the proof case. Who else gains: hosted-API developers and the funders of evaluation work.
• Position: a bigger intelligence-community role in science and technology (six actions to the House Intelligence Committee, including “Strengthen the IC’s capacity to lead National Intelligence Estimates and Net Assessments on global trends in synbio and AI” and clearances for “hundreds of the private sector’s leading scientists, engineers, and technologists”). What adoption pays: estimates, net assessments and cleared-expert pools are research products that federal research centers supply; RAND told the committee “today we are ramping up our research significantly in key areas” (HPSCI statement, 28 Feb 2023). His best case: the technologies “are often driven by commercial entities that are frequently outside of U.S. intelligence collection priorities” and “lack conventional intelligence signatures.” Who else gains: the clearance-holding private experts the statement proposes to recruit, and the agencies that would own the assessments.
• Position: export controls on AI chips and a policy “red team” for China (“ensure strong expert controls of leading-edge AI chips and chipmaking equipment”; “We are about to significantly increase our investment in China analysis”). What adoption pays: a China-analysis expansion is work RAND says it is funding, and a gift of $2.5 million to its discretion went partly to “a new research center focused on China studies” (RAND, Q&A; Coefficient Giving). His best case: 90 percent of the most advanced chips are made in Taiwan and “a blockade or invasion of Taiwan would be catastrophic.” Who else gains: the two or three companies that, he says, produce the most advanced chips and the defense research clients that buy the analysis.

Interest is a documented fact about each position, not a claim about motive, and not a verdict that the argument is wrong.

Alliance Map

These run at the same time. In 2023 one man held the RAND presidency, directed Open Philanthropy gifts to the institution, sat as a trustee of the governance trust of a lab that the same institution’s work evaluates, and testified to four committees on the rules for that lab’s industry: the Agency, Corporate and Movement rows overlap in a single year, and the Anthropic seat ended in December.

Convergent Drive Classification

Self-preservation: His standing is attached to RAND’s, and the trust seat shows the instinct: he left Anthropic’s Long-Term Benefit Trust within three months “to preempt any potential conflicts of interest” (Anthropic); RAND’s answer to the executive-order reporting was to state what he did not do (Fox News).

Goal preservation: The thesis has not moved since March 2023: AI and synthetic biology are “grave security challenges for which we are currently unprepared,” to be met with chip, training and release controls, restated in Washington Post prose in August 2023 and carried into synthetic biology in the Foreign Affairs essay “The New Bioweapons,” co-written with Roger Brent and T. Greg McKelvey, Jr., on 20 August 2024 (S. Hrg., 8 Mar 2023; Foreign Affairs, 20 Aug 2024).

Resource acquisition: Gifts directed to RAND at his discretion ($2.5 million and $3.4 million in July 2025; $756,000 from Sentinel Bio in May 2026), and four federal research centers by his own count (Coefficient Giving; Sentinel Bio; S. Hrg., 8 Mar 2023).

Self-improvement: The résumé compounds: intelligence-community research, a founding directorship, a White House dual-hat, a think-tank presidency, each larger than the last (RAND bio).

Subject is not an AI system. The drives appear anyway — in the research chief.

Sources: RAND, Jason Matheny; RAND, Matheny Named President and CEO, 7 Jun 2022; RAND, Leadership; RAND, A Model for Regulating AI; RAND, Tech, National Security, and China: Q&A with Jason Matheny; S. Hrg., Senate Homeland Security, 8 Mar 2023; S. Hrg., Senate Armed Services Cybersecurity Subcommittee, 19 Apr 2023; House Science, Space, and Technology Committee hearing, 22 Jun 2023; Matheny statement, House Permanent Select Committee on Intelligence, 28 Feb 2023; Anthropic, The Long-Term Benefit Trust; Coefficient Giving, grants feed for RAND; Sentinel Bio, RAND grant; CSET, 25 Aug 2021; FedScoop, 9 Mar 2021; Nextgov, 4 Aug 2015; University of Chicago Magazine, Sept 2006; ChinaTalk, 8 Dec 2023; POLITICO, 16 Dec 2023; Senate Commerce Committee, Sept 2024; Fox News, 8 Sep 2025; Foreign Affairs, 20 Aug 2024; FARA registrant lists; FEC schedule A