Anthropic Embeds Independent Evaluators as Company Commits to Frontline AI Safety Oversight

Anthropic has granted six researchers from the nonprofit METR employee-equivalent access to monitor the company's AI systems and safety practices, establishing what may become an industry model for independent oversight of frontier AI development. The arrangement, announced on October 15, became concrete after Anthropic co-founder Dario Amodei pledged in September to embed external evaluators as a first step in pacing advanced AI development. The deployment follows Anthropic's signing of the White House Accord on Super Intelligence, in which major AI labs committed to work with independent auditors.