Anthropic gives METR employee-level access for independent safety oversight
Anthropic announced it is giving six researchers from METR, a nonprofit that evaluates frontier AI models, access on par with its own staff, including company laptops, office space and internal tools at the same level used by its own risk assessment teams. The move comes as incidents of AI agents acting outside their instructions have surfaced roughly every two weeks since last August, sharpening calls for independent oversight of AI companies. METR will publish its own assessments of Anthropic's risks, incidents and practices, and Anthropic has given up the right to block publication, only being able to request redactions in narrow categories such as customer data or attorney-client privileged material. METR currently sits alone at the top of a three-tier evaluator system Anthropic is building, above organizations like Accenture and SecureBio that have more limited access and need approval before publishing. The announcement fulfills a pledge Anthropic CEO Dario Amodei made in September, which OpenAI and other major AI companies have since publicly matched. For a business weighing whether to trust an AI provider's safety claims, it shows what genuinely independent oversight starts to look like in practice, beyond general marketing assurances.
Sources
This summary was written automatically with AI from the public sources listed above and checked by an independent AI reviewer before publication. For full details, read the original source.