Anthropic gives METR researchers employee-level access to review its AI risks

Anthropic is granting researchers at METR employee-level access to its internal systems. METR plans to publish its own independent assessments of Anthropic's risks, incidents, and practices. Under the arrangement, Anthropic has relinquished any right to block or veto those publications. The move gives an outside research group a deeper look into how a frontier AI developer operates. The source article provides no further detail on timing or specific findings.