Anthropic CEO Calls for Slower AI Pace, Opens Models to External Review

Anthropic chief Dario Amodei announced that the company will slow its AI development and grant third‑party evaluators, such as METR, broad access to its models. In a public statement, Amodei outlined a three‑step plan to "pace the frontier": first, reduce the speed of training and new model releases; second, allow independent reviewers to assess safety and compliance; third, bring the wider industry and regulators together to establish shared safeguards. The move is presented as a unilateral step toward greater transparency and a commitment to safety practices. Anthropic’s decision follows growing calls for tighter oversight of large language models and aims to give regulators time to evaluate emerging technologies before they are deployed.