Anthropic's proposed AI watchdog METR has deep ties to Effective Altruism

This reports on Anthropic CEO Dario Amodei’s proposal to embed third-party evaluators from a nonprofit called Model Evaluation and Threat Research (METR) inside AI companies to monitor safety, review training pipelines, verify commitments and report incidents. METR spun out of an Effective Altruism-aligned incubator, Alignment Research Center, and its leadership and staff include figures tied to that community: founder/CEO Beth Barnes, chief scientist Hjalmar Wijk, staffers Ajeya Cotra, Megan Kinniment and Pip Arnott, and links to Paul Christiano. METR says it won’t take funding from AI firms and requires disclosures of conflicts, but publicly reported backing includes donors aligned with EA and Anthropic investors such as Dustin Moskovitz and Jaan Tallinn; METR has raised tens of millions and announced $71 million in recent funding.

Critics assert those overlaps create serious conflicts of interest, calling METR effectively an Anthropic proxy and questioning its independence, while supporters argue embedded auditors would increase transparency. Political and regulatory pushback is already evident: President Trump and other conservatives dismiss AI doomsday claims, FTC officials warn against industry self-regulation and antitrust exemptions, and commentators note potential product‑liability motives for Amodei’s pace-the-frontier warnings. The proposal follows internal departures and high-profile warnings about rapid AI deployment, making METR’s role a focal point in debates over who should police advanced AI.