October 15th 2026, 8:46 PM
Anthropic opens its AI safety lab to outside evaluators
Key Summary
Anthropic is expanding its work with independent safety researchers by granting access to its AI models and internal systems to organizations like METR and SecureBio. This move aims to promote safety and transparency in AI development, with a focus on evaluating risks associated with its Claude models. While the arrangement provides significant access, Anthropic may make limited redactions for sensitive information, leaving the default as disclosure rather than approval.
Please see our real time news feed on our Home Page
Introduction
Anthropic is letting outside safety researchers into the building.The New Arrangement
On October 15, 2026, the AI developer expanded its work with independent evaluators. A six-member team from the nonprofit METR (Model Evaluation and Threat Research) receives what amounts to employee-like access to the company's hardware, office space and internal systems.SecureBio and Accenture's Faculty unit also hold defined roles in a new three-tier evaluator framework.
How the Three Tiers Work
SecureBio operates at tier one. Accenture's Faculty unit handles tier two.Why Independent Evaluation Matters
The arrangement ties back to commitments in Anthropic's Advanced AI Framework.What This Means
The framework is only as meaningful as what comes out of it.Conclusion
The redaction clause deserves close attention.Biosecurity Remains the Area to Watch
SecureBio's review of the August 2026 report will show whether its earlier assessments hold for newer work.#AI#Safety#Regulation#Anthropic#METR