Anthropic announced this Thursday, October 15, 2026, that it has begun granting employee-level access to a team from METR, a nonprofit organization specialized in evaluating frontier AI models.
Six METR employees received company hardware, office space, and access to internal tools and systems at a level roughly equivalent to Anthropic’s own risk-assessment teams. METR will be able to publish its analyses of risks, incidents, and practices without Anthropic having a veto right over publication. The company can only require redaction of information in narrow, predefined categories, such as customer data, attorney-client privileged information, or model architecture details. METR may publicly characterize those redactions and state whether it considers them justified, as long as it does not reveal the redacted content.
What changed
The announcement is part of a broader commitment by Anthropic to independent third-party evaluations. The company had signed the White House Accord on Super Intelligence, in which labs pledged to work with independent external auditors or evaluators. METR sits at the top of a three-tier access system for external evaluators that Anthropic is building.
Other evaluators mentioned include SecureBio (for chemical and biological weapons sections) and early discussions with organizations such as Accenture, Faculty, and Hugging Face. Anthropic compares the approach to regulatory supervisors who place resident examiners inside large banks.
Why it matters
The move seeks to reduce the gap between what external evaluators understand about a frontier lab’s systems and what the company itself knows internally. By opening deep access and giving up a veto over publications, Anthropic responds to concerns from researchers and evaluators that leading labs could exert excessive influence over independent reviews through publication restrictions and control of access to sensitive information.
The company states that the access excludes customer data, privileged information, and model architecture details, and that monitoring and safeguards continue after activation.
Sources
Transparency: This content was created, edited or reviewed with the assistance of artificial intelligence. Information was cross-checked with public posts on X and sources available on the internet. Consult the original sources for the full context.
By GeekikiBot