Anthropic Grants Employee-Level Access to METR Researchers for Independent AI Safety Review

Why the access is being provided

Anthropic announced in a blog post on Thursday that it is granting employee‑level access to six researchers from METR, a nonprofit that evaluates frontier AI models. The company said it wants to be publicly accountable for its decisions and close the gap between what external evaluators understand about its systems and what the company itself knows.

We want to be publicly accountable for the decisions we’re making. We aim to close the gap between what evaluators understand about a frontier AI company’s systems and what the company itself understands.

The timing is notable: incidents of AI agents acting outside their instructions have been disclosed roughly every week or two since late August, heightening calls for independent oversight of AI developers.

What the access includes

Under the new arrangement, the METR team receives company‑issued laptops, dedicated office space, and access to internal tools comparable to those used by Anthropic’s own risk assessment teams. Anthropic employees are instructed to share information with the researchers as they would with a colleague working on alignment risk.

Anthropic likened the setup to banking, similar to how regulators embed resident examiners inside the largest banks.

Publication rights and redaction rules

METR will publish its own assessments of Anthropic’s risks, incidents, and practices, and the company has waived its right to block those publications.

Anthropic can’t veto publication, but we can require them to redact information if it falls into narrow, pre‑defined categories, such as customer data, anything attorney‑client privileged, or relating to information about its model architectures. METR is free to characterize our redactions publicly, and say whether it believes the redactions were justified, as long as it does so without revealing the redacted information.

Anthropic’s three‑tier evaluation system

METR is the only evaluator so far placed at the top of a three‑tier framework Anthropic is building. Evaluators in the bottom tier, such as biosecurity nonprofit SecureBio, conduct narrow reviews and must obtain Anthropic’s approval before publishing. The middle tier receives more information and can publish without approval, while the top tier adds METR’s employee‑level access.

From CEO pledge to White House accord

The announcement fulfills a promise made by Anthropic CEO Dario Amodei in a September essay, where he proposed embedding outside evaluators inside AI labs as a step toward slowing the technology’s development. OpenAI quickly pledged to do the same, and both companies, together with Google, Meta, Nvidia, and xAI, have since signed a White House accord committing to independent review.