Anthropic grants METR employee-level access for independent safety evaluations

Anthropic announced it is granting six researchers from METR, a nonprofit focused on testing frontier AI models, employee-level access to independently assess its models, safety practices, and risk management. The researchers have been issued Anthropic company laptops, office space, and access to internal tools comparable to those used by Anthropic’s own risk assessment teams.

Under the arrangement, Anthropic employees are encouraged to share information with METR researchers as they would with colleagues working on alignment risk. Anthropic compared the setup to regulatory practices in banking, where examiners are embedded within institutions to oversee compliance and risk.

METR will publish its own assessments of Anthropic’s risks, incidents, and practices, and Anthropic has relinquished its right to block publication. The company may only require redactions for narrowly defined categories, such as customer data, attorney-client privileged information, or details about model architectures. METR is permitted to publicly characterize these redactions and state whether it believes they were justified, provided it does not reveal the redacted content.

The announcement formalizes a three-tier evaluation system at Anthropic. Evaluators in the lowest tier, such as the biosecurity nonprofit SecureBio, conduct narrow reviews and require Anthropic’s approval to publish. The middle tier allows for publication without approval, while the top tier—represented by METR—includes the highest level of access and independence.

Anthropic’s move follows a September essay by CEO Dario Amodei proposing embedding outside evaluators inside AI labs as a step toward slowing frontier AI development. OpenAI subsequently pledged to adopt a similar approach, and both companies, along with Google, Meta, Nvidia, and xAI, signed a White House accord committing to independent review.