METR Clarifies Mission Amid AI Evaluator Independence Debate

Episode 1 · METR President Responds to Critics, Reaffirms AI Risk Evaluation Mission (2026-09-17, 6 posts)

METR President Chris Painter personally published a post around September 17 responding to recent criticism of the organization and its OpenAI audit, reaffirming METR's mission and positioning. Multiple bloggers (@AndyMasley, @basedjensen, @dhadfieldmenell, @kyliebytes) relayed the piece, and researchers including Talia Ringer shared it in support. The upshot: Painter frames METR as a research organization evaluating AI risks, not a regulator, but questions about its actual access during the OpenAI audit remain unclarified by first-hand sources.

Confirmed

• Painter stated that since 2022 METR has worked to ensure the public is informed "if AI truly becomes autonomous, incorrigible, or close to out of control," positioning METR as an evaluator of loss-of-control risk rather than a regulator.
• He noted team members gave up high-paying jobs for AI safety evaluation, and that METR has cooperated with frontier labs including OpenAI since 2022.
• Context includes Ben Todd joking that "METR's biggest donor is the mysterious TED talk organization."

Unconfirmed

• @NathanpmYoung relayed claims that METR allegedly received only a 3-day information access period in its OpenAI investigation and was barred from materials outside that window; citing MackenZarnold, he argued Stoller's claim of a "sham investigation" reflects a forced compromise rather than negligence. These details come second-hand and the contract terms remain unverified.

Why it matters

• METR is one of the few third-party evaluators of dangerous AI capabilities; its independence and actual access directly affect the credibility of frontier-model safety assessments. If access was indeed limited, the public must recalibrate what its reports do and do not cover.

• METR President Clarifies Mission: Surface AI Loss-of-Control Evidence, Not Regulate — kyliebytes · 2026-09-17
• METR president fires back at critics: staff forgo high pay for AI safety evals — AndyMasley · 2026-09-17
• METR president re-ups mission: surface evidence if AI nears going rogue — dhadfieldmenell · 2026-09-17
• METR president Chris Painter explains what the AI eval org actually does — dhadfieldmenell · 2026-09-17
• METR's OpenAI probe got just 3 days of access — defenders say supplicant status, not incompetence — NathanpmYoung · 2026-09-18
• METR President Explains Mission: Making AI 'Going Rogue' Evidence Public — basedjensen · 2026-09-18

Episode 2 · AI Safety Evaluator METR Discloses Funders and Conflict-of-Interest Policy (2026-09-17, 3 posts)

METR has disclosed its funders, including The Audacious Project and Schmidt Sciences, published its conflict-of-interest policy, and confirmed it refuses donations from frontier AI companies.

• METR Discloses Its Funders, from Audacious Project to Schmidt Sciences and Dylan Field — CFGeek · 2026-09-17
• METR publishes its conflict-of-interest policy and funder information — CFGeek · 2026-09-17
• METR lands TED Audacious Project funding, refuses money from frontier AI labs — ben_j_todd · 2026-09-17

Episode 3 · Scholars debate who should run independent AI model evaluations (2026-09-17, 3 posts)

Yoav Artzi and others debate the incentives of AI evaluation bodies, arguing universities face funding conflicts but are cleaner than lab-affiliated nonprofits, and that evaluations should be done by stable, well-paid career researchers rather than low-paid students.

• Debate: universities face fewer incentive conflicts than lab-connected nonprofits for AI evals — tallinzen · 2026-09-17
• AI evals shouldn't be done by underpaid students but by stable, tenured research staff — tallinzen · 2026-09-17
• Prof Yoav Artzi: universities doing AI evals risk losing lab funding and face liability — yoavartzi · 2026-09-17