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

Anthropic gave six staffers from the nonprofit METR inside access to its AI models and safety practices, including company hardware, office space, and internal tools, putting them roughly on par with its own risk teams.

• Customer data, attorney-client privileged information, and model architecture details remain off-limits, and METR will publish its findings independently.
• Anthropic says the arrangement goes further than access offered to other outside reviewers such as SecureBio, Accenture, and Faculty.