Anthropic CEO Dario Amodei calls for slowing AI development to ...

Dario Amodei, the chief executive of Anthropic, has proposed a three-step plan for managing the risks of artificial intelligence, according to a report published by jpost.com. Amodei called the approach "pacing the frontier," and the report's headline states that he called for slowing AI development. The proposal consists of three named components set out in sequence: Embedded Evaluators, Democratic Coordination, and Global Coordination. The report presents the plan as Amodei's proposal rather than as a policy adopted by Anthropic, by any government, or by any international body.

The first component, Embedded Evaluators, would give independent third parties access to assess safety practices, according to the report. The second, Democratic Coordination, would have AI companies in democratic countries establish shared safety standards. The third, Global Coordination, would have governments work together to address AI risks while verifying compliance. The report describes the plan as a proposal; it does not state that any company or government has adopted it.

The report states that Amodei argues "pacing the frontier" must serve a practical purpose. He argues that the extra time created by pacing should be used to make AI development safer. He further argues that the approach should not become an empty exercise — that is, a slowdown that achieves nothing. The report does not state a timeline for the plan, an enforcement mechanism, or any party that has agreed to take part.