Anthropic CEO Calls for AI Industry to Slow Capability Advances

Anthropic CEO Dario Amodei has called on the artificial intelligence industry to slow the pace of its capability advancements to allow safety measures to catch up, outlining a framework for embedded third-party evaluators and coordinated safety standards. In an essay published on Saturday, Amodei argued that the rapid progress in AI, driven by models' growing ability to improve themselves, is outstripping the industry's ability to understand and control these systems. "We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain," Amodei wrote. He emphasized he was not calling for a halt to training or technical progress, but for a deliberate pacing to ensure adequate time for alignment and safeguarding. Amodei's proposal centers on permanently embedding independent third-party evaluators inside frontier AI companies. These auditors would have ongoing access to internal systems and risk-assessment processes to monitor safety practices and report incidents. Anthropic has committed unilaterally to this measure, granting permanent access to teams like METR.