Anthropic Gives Independent AI Safety Researchers Employee-Level Access to Internal Systems
Anthropic is giving six METR researchers employee-level access to internal tools, laptops and office space to assess models and safety practices. METR can publish findings without a company veto, though narrow redactions cover customer data, privileged material and model architectures.
Anthropic announced in October 2026 that it is granting six researchers from METR, a nonprofit that tests frontier AI models, employee-level access to independently assess its models, safety practices, and risk management.
The METR team has been issued Anthropic company laptops, office space, and access to internal tools comparable to those used by Anthropic’s own risk assessment teams. Anthropic employees are encouraged to share information with the METR researchers as they would with a fellow employee working on alignment risk, the company said.
Critically, Anthropic has given up the right to block METR from publishing its findings. “Anthropic can’t veto publication, but we can require them to redact information if it falls into narrow, pre-defined categories,” the company said, citing customer data, attorney-client privileged material, and information about model architectures as examples. METR is permitted to publicly characterize any redactions and state whether it believes they were justified.
The arrangement is part of a three-tier evaluator system Anthropic is building. METR sits at the top tier with full employee-level access. A middle tier allows evaluators to publish without Anthropic’s approval but with less access, while a bottom tier — which includes biosecurity nonprofit SecureBio — covers narrower reviews and requires Anthropic’s approval to publish.
The move comes amid a period of heightened scrutiny of AI companies. Incidents involving AI agents acting outside their instructions have been disclosed roughly every week or two since late August 2026, intensifying calls for independent oversight. Anthropic compared the METR arrangement to how regulators place resident examiners inside major banks.
The announcement fulfills a commitment made by Anthropic CEO Dario Amodei in a September essay, in which he proposed embedding outside evaluators inside AI labs. OpenAI subsequently pledged the same, and both companies, along with Google, Meta, Nvidia, and xAI, have since signed a White House accord committing to independent review.