Anthropic gives METR inside access to audit its AI models and safety practices

Anthropic has given a team from the nonprofit METR inside access to review its AI models and safety practices, the company announced. Six METR staffers received company hardware, office space and access to internal tools, which Anthropic describes as putting them roughly on par with its own internal risk teams.
The arrangement comes with defined boundaries. Customer data, material covered by attorney-client privilege, and details about model architecture remain off-limits to the reviewers. METR will publish its findings independently, and Anthropic says it may request redactions in certain predefined categories. METR retains the freedom to state publicly whether it believes those redactions were justified.
Anthropic's stated rationale is that not all safety-relevant information can be released, either because disclosure could help attackers or because it is commercially sensitive. The METR deal goes further than access the company has offered other outside reviewers: SecureBio, Accenture and Faculty have so far received less access and tighter limits on what they can publish. Anthropic has also started early talks with Hugging Face about a similar arrangement. The move follows Anthropic's signing of the White House Accord on Super Intelligence, which includes a commitment to working with outside reviewers.
Why it matters: independent audit access is a structural change in how frontier AI safety claims can be checked, and it sets a precedent other labs may be pressed to match. For developers and product teams, it signals that third-party scrutiny of safety practices is becoming part of the release landscape, though the carve-outs for architecture and customer data mean the deepest technical details stay private. Whether METR's independent publication proves substantively critical will be the real test of how much this access is worth.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Anthropic has given a team from the nonprofit METR inside access to review its AI models and safety practices, the company announced. The article Anthropic gives METR inside access to audit its AI models and safety practices appeared first on The Decoder .