Anthropic announced on Thursday that it is granting employee-level access to six researchers from METR, a nonprofit that tests frontier AI models. According to the company, the arrangement allows METR researchers to independently assess Anthropic's models, safety practices, and risk management. The METR team has received company-issued laptops, office space, and internal tool access comparable to Anthropic's own risk assessment teams. Anthropic compared the setup to how regulators place resident examiners inside major banks. Under the agreement, METR will publish its own assessments of Anthropic's risks, incidents, and practices. Anthropic has given up the right to block these publications, though it can require redactions for narrow categories such as customer data, attorney-client privileged information, or model architectures. METR is permitted to publicly characterize any redactions and state whether it believes they were justified. This initiative places METR at the top of a new three-tier evaluation system being built by Anthropic, which also includes narrower reviews for bottom-tier organizations like SecureBio. The announcement fulfills a proposal made by Anthropic CEO Dario Amodei in a September essay suggesting the placement of outside evaluators inside AI labs. The move comes amid a growing frequency of disclosures regarding AI agents acting outside their instructions, which has amplified calls for independent oversight across the industry.
Anthropic is letting METR researchers look under the hood with employee-level access
Anthropic has granted six researchers from the nonprofit METR employee-level access to independently assess its frontier AI models and safety practices. METR will publish its own risk assessments without Anthropic holding veto power over publication, subject to narrow redactions. The arrangement follows recent incidents of AI agents acting outside instructions and fulfills a proposal made by CEO Dario Amodei in September.
Why it matters
This unprecedented level of insider access establishes a new benchmark for independent oversight and public accountability among frontier AI developers. It answers growing calls for external checks as AI models become more autonomous.