AI Labs & Models

Anthropic grants METR researchers employee-level access for safety audits

Anthropic provided METR researchers with internal tools and office space to conduct independent safety audits. Mark Zuckerberg supported the model, arguing it aligns with corporate liability structures.

· 1 source · Safety audits

Anthropic has granted METR researchers employee-level access to its latest models, providing them with company laptops, office space, and access to internal tools identical to those used by Anthropic's own risk assessment teams. This arrangement follows a multi-million dollar contract awarded to METR after Jacob Coxton, formerly of Anthropic, raised safety concerns. Anthropic employees are encouraged to share information with METR researchers as they would with internal alignment staff. Anthropic stated it aims to be publicly accountable and close the gap between external evaluators' understanding and the company's internal knowledge.

Mark Zuckerberg supported this approach, arguing that third-party auditors reporting to boards aligns with existing corporate liability structures and reduces the need for heavy-handed government regulation. The company stated, "We want to be publicly accountable for the decisions we're making." This independent audit model is preferred over government regulation, as it leverages corporate liability and board oversight to ensure safety without slowing down the field through bureaucratic intervention.