Anthropic has granted researchers from METR (formerly known as Epochal) employee-level access to its systems, enabling comprehensive independent audits of the company's AI safety practices, incident handling, and risk management. Critically, Anthropic has relinquished its right to block or suppress METR's published findings.

This arrangement represents a transparent approach to third-party oversight of frontier AI development. METR's team will publish independent assessments of Anthropic's practices and any security vulnerabilities or incidents they discover, without Anthropic having editorial control.

What This Means for Your Business

Organizations evaluating AI vendors should note this model of independent verification. Anthropic's willingness to submit to uncontrolled external audits signals confidence in its safety practices and sets a new standard for AI company accountability. For enterprises, this level of transparency can reduce procurement risk and provide audit trails for regulatory compliance requirements around AI governance.