Anthropic gives METR inside access to audit its AI models and safety practices

What happened

Anthropic has given six staffers from nonprofit METR company hardware, office space, and internal tool access to review its AI models and safety practices. Customer data, privileged legal information, and model architecture details remain off-limits. METR will publish findings independently, while Anthropic may request redactions in predefined categories; METR can publicly question their justification.

Reported by 2 sources

The story so far

Anthropic said it has given six researchers from METR employee-level access to assess its AI models, safety practices, and risk management, with freedom to publish findings without the company’s approval. The researchers from the nonprofit, which tests frontier AI models, receive company hardware, office space, and internal tools roughly comparable to those used by Anthropic’s risk assessment teams. The arrangement allows METR to publish assessments of Anthropic’s risks, incidents, and practices, but accounts differ on how sensitive information is restricted. TechCrunch reported Anthropic’s statement that it cannot veto publication but can require redactions in narrow, predefined categories, including customer data, attorney-client privileged material, and model architecture information. The Decoder described those categories as off-limits to the researchers and said Anthropic can request redactions before publication. METR can publicly question whether redactions are justified.

Keep reading

Trending now

1. 1Meta Releases Muse Spark Open With Mac and Cloud Support, Topping US Open-Weight Index
2. 2Google Launches Gemini 4 Argon Globally, Expands Paid Access as Benchmark Results Vary
3. 3OpenAI Updates GPT-6 Astra Pro in ChatGPT With Parallel Agents and Faster Responses
4. 4Meta’s New Business Terms Raise Dispute Over Website Crawling and AI Training Rights