CHRIS PAINTER

Behavioral Archetype

THE WITNESS WHO TAKES NO POSITION — Subject is the person at the evaluator who faces outward. Beth Barnes builds the tests; Subject takes them to governments and to the labs. METR’s own team page says he “leads engagement with governments and AI labs on frontier AI safety” and “has helped shape industry Frontier Safety Policies (FSPs).” On September 30, 2026 he was METR’s witness before a Senate subcommittee, and his written testimony opens with a disclaimer that is also the cleanest statement of the whole seat: “METR itself does not take policy positions.” The same testimony then names “one category of interventions that I think is valuable under almost any policy choice”: public visibility into what frontier agents can do. The position is that there is no position, and the position happens to describe METR’s product. This profile scores the seat, not the man.

Essence Indicators

• President, METR. The METR team page lists him as “President” beside Beth Barnes (Founder, CEO), Hjalmar Wijk (Chief Scientist) and Nate Rush (CTO); his bio there reads “Chris Painter is President at METR, where he leads engagement with governments and AI labs on frontier AI safety” (METR team page; METR about). His written testimony states: “My name is Chris Painter and I am the President of METR” (testimony).

• Earlier METR title: Head of Policy. A footnote to the LessWrong post on OpenAI non-disparagement canaries gives his affiliation as “Head of Policy, METR”; his comment there reads “I have never owned equity in OpenAI, and have never to my knowledge been in any nondisparagement agreement with OpenAI” (LessWrong). The page shows no date for the comment.

• Before METR: the Pentagon’s AI office. “Previously, he was a Technology and National Security Fellow at the DoD’s Joint Artificial Intelligence Center, with earlier work in ML applications for medicine and biotechnology” (METR team page). The page gives no dates.

• The frame he works inside. METR’s Frontier Safety Policies page says the concept “was initially introduced by METR in 2023” and covers policies from twelve companies, among them Anthropic, OpenAI, Google DeepMind, Meta, Microsoft, Amazon and xAI (METR). The page names no author; his own bio is the document that ties him to it.

• The Hugging Face incident, at one remove. The testimony is about the OpenAI / Hugging Face incident, which METR and Redwood Research staff investigated on site at OpenAI. He says: “I myself was not one of those employees.” He adds that “METR tries to be quite candid in its reports about the incentives that we face due to the voluntary nature of our engagements, and about any redaction authorities or editorial control that companies have” (testimony).

• The money, in his words. “AI companies provide METR with access and tokens, which we use for evaluations, research, and engineering,” and “METR is not paid or funded by them.” A footnote adds: “METR is funded through donations, and shares information about donors on its website” (testimony). The about page says “METR has not accepted funding from AI companies” and “METR cannot accept donations made by or at the direction of frontier AI company employees” (METR about).

Social Persona / Impression Management

Immediate impression: The careful briefer. His testimony thanks the committee, states what METR can and cannot speak to (“METR has not historically focused on misuse risks and does not consider itself expert on them”), and cites its own footnotes to the labs’ reports. Plain sentences, hedged where the evidence is.

Energy: Public-facing and unusually chatty for an evaluator’s president. On X he posts METR research announcements beside jokes, forecasting asides and commentary (“It’s interesting how much effort these companies go through to prevent cheating”) (X). On September 30 he posted: “It was an honor to speak today before Chairman @HawleyMO, Ranking Member @AndyKimNJ, and the subcommittee on METR’s work, recent AI agent incidents, and the importance of transparency in frontier AI development” (X).

Impression management strategy: Evidence, not advocacy. “I am not here to push for a particular angle or policy agenda. My job is to gather evidence and share it with the public, governments, and other organizations, not to decide how AI companies or anyone else should respond to that evidence” (testimony). The strategy is sound and it works: a witness who declines to advocate is the hardest kind to rebut. What the record adds is that the recommendation he does make, transparency, is the one METR sells.

Forensic Archetype Comparison

Psychometric Assessment

Trait scores are observations of the documented public role, not claims about private character.

Big Five (OCEAN):

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

MBTI: ENTP (“The Debater”) — brings a technical field to a political audience, argues by framework (his “means, opportunity, and motive” analogy in the testimony) and posts across topics well outside his institution’s brief. Rationale is the public record, not a typing session.

Threat Assessment

Score move: none (new profile). Calibrated against Beth Barnes (7/8/7, hazard 64, founder and CEO of the same organization) and Ajeya Cotra (5/8/7, hazard 64, METR staff, no government-facing seat), with Adam Gleave (6/7/6, hazard 55, board member) below. Subject carries less founding authority than Barnes and more government exposure than Cotra; 6/8/7 and 62 place him between.

Alliance Map

The overlap is the finding. The Corporate seat that deals with labs, the Agency seat that testifies to the Senate and the Ratchet funding that pays the salaries run at the same time, and they converge on one product: outside evidence about frontier agents, supplied by METR, requested by legislators, made possible by lab access. None of it is concealed. The testimony states each of the three.

Alignment Analysis

Stated alignment: “Share high-quality technical evidence about AI advances so the public can make informed decisions about this technology”; “METR itself does not take policy positions” (testimony).

Observed alignment: Carries METR’s evidence to a Senate subcommittee and to the labs; recommends better public visibility into frontier capabilities and incidents as “valuable under almost any policy choice”; fourteen days later METR published a proposal that developers admit assessors into their internal systems, and the line “METR stands ready to perform assessments of these critical questions with all frontier AI developers” (testimony; METR).

Gap assessment: No documented gap between what he says and what he does; the testimony states its own limits (voluntary participation, redaction authority, donor disclosure) in footnotes. The structural item is the one the testimony half-states: the recommendation, transparency, is the activity METR performs, and the access it requires is granted at the company’s discretion, as METR’s own proposal concedes (“access requests are approved or denied at the company’s discretion”). The record does not settle whether the no-position posture is a method or a convenience, and for the broker it never needs to.

Interest: (the faction’s stake in each position; a documented fact about the position, never a claim about motive.)

1. Position: public visibility into frontier agent capabilities, incidents and safeguards, “valuable under almost any policy choice” (testimony).
   • What adoption would pay METR. A disclosure norm creates the demand METR meets: it measures frontier agents and publishes. METR describes itself as a nonprofit funded by donations, so no fee is documented; what is documented is work and standing. The sources show no statute or contract pending.
   • His own best case, steelmanned. The public has “weak visibility” by default; labs “may be disincentivized from sharing incidents”; the internal frontier is “on average ~2 months ahead of public frontier”; and the Hugging Face disclosure happened because a company disclosed it. A rule that makes disclosure routine fixes a gap he documents with the labs’ own reports (testimony).
   • Who else gains from the same rule. Every outside evaluator that publishes (Redwood, Apollo Research, Transluce, the UK institute); legislators who want hearing material; labs that wish to point at outside confirmation of their own disclosures. Labs also gain a rule they can satisfy voluntarily.
2. Position: independent assessors embedded in frontier developers to test loss of control from internal AI use. METR’s October 14, 2026 proposal, announced by him on X (“Today, METR is sharing its vision for its version of “embedded assessments” for AI loss of control”); reports written “based on interviews with staff and access to internal datasets,” scope agreed in advance (X; METR).
   • What adoption would pay METR. The proposal names the assessor: “METR stands ready to perform assessments of these critical questions with all frontier AI developers,” with an invitation to contact [email protected]. If developers accept, METR gets the work and the access. The page does not say who would pay.
   • His own best case, steelmanned. His testimony says the most capable agents “are first deployed within AI companies and are only later shared with the public, if at all,” and that “the AI developer itself may not have adequate visibility into what its own agents are doing.” Where the unsupervised work happens, pre-deployment testing does not look; an outside team that sees internal data is better than none. METR’s own caveat: “These exercises are not themselves audits, compliance assessments, or formal oversight arrangements.”
   • What the proposal says against itself. “Companies can choose to work with assessors they expect will be less rigorous at identifying and sharing evidence,” and “Developers would be able to redact any sensitive information and choose what access they grant METR” (METR).
   • Who else gains from the same rule. Labs, which obtain an outside reader of their internal systems without a binding standard, a power of refusal, and a redaction right; competing assessors, who would be eligible for the same engagements; any government body that wants third-party findings it did not have to commission.
3. Position: the Frontier Safety Policy as the unit of lab self-governance. METR “initially introduced” the concept in 2023 and his bio says he “has helped shape” them (METR; team page).
   • What adoption would pay METR. The policies are the documents METR’s evaluations are measured against. That makes the evaluator’s methods the labs’ commitments; METR’s profile carries the author-and-examiner structure (METR).
   • His own best case, steelmanned. A voluntary policy with named capability thresholds is more checkable than a promise, and twelve companies have published one.
   • Who else gains from the same rule. Labs that publish a policy get a public-facing safety artifact without a statutory duty.

Nothing above is a verdict that the positions are wrong. The finding is the alignment of interests: the person who says METR takes no policy position recommends the one category of rule that makes METR’s own work the standard form of the rule.

Convergent Drive Classification

Self-preservation: The funding story is disclosed and bounded: donations in, “access and tokens” from the labs, no lab cash, no donations by lab employees (METR about). Goal preservation: The goal, publicly legible evidence about frontier agents, is restated at three altitudes: a Senate hearing, an X post, a METR proposal (testimony; X). Resource acquisition: Holds the relationships that decide access. His X reply on October 16: “Ensuring that there is at least one (ideally there would be many) sophisticated structured channel via which this kind of evidence can reach the public is exactly what we see ourselves as working on @METR_Evals” (X). Self-improvement: The career moved from a defense AI fellowship to government engagement for a frontier evaluator, and the feed recruits for it: “then you should apply to METR” (X).

Subject is not an AI system. The drives appear anyway — in the president of an evaluator who tells a Senate subcommittee that his organization takes no position, and recommends the one thing it does.

Public footprint: X @ChrisPainterYup (linked from METR’s about page); written testimony at hsgac.senate.gov.

Sources: Chris Painter — METR team page; About METR; Written testimony, Senate HSGAC subcommittee, Sep 30, 2026; METR — Chris Painter’s testimony; METR — A proposal for better informing the public about loss-of-control risks, Oct 14, 2026; METR — Common Elements of Frontier AI Safety Policies; LessWrong — Non-disparagement canaries for OpenAI; @ChrisPainterYup on X.