Anthropic Grants METR Employee-Level Access for Independent AI Safety Audits

16 Oct 2026, 12:44 am · 2h ago · 1 min read · TechCrunch AINew

Frontier AI lab Anthropic has granted independent research organization METR employee-level access to evaluate its internal AI safety practices, risks, and potential incidents. Under the terms of the agreement, METR has the authority to independently publish its third-party findings and risk assessments without Anthropic holding any veto or blocking rights over the disclosures. This arrangement marks an unprecedented level of external transparency for a major frontier model developer. The initiative sets a new standard for AI governance and third-party safety audits amid mounting global regulatory scrutiny surrounding advanced generative AI systems.