AI Governance
Anthropic gives METR employee-like access to independently evaluate frontier model risks
Anthropic on October 15, 2026 said six METR researchers now have near-employee access to internal systems to assess models and safeguards, with freedom to publish risk findings, while also naming SecureBio and Accenture …