Why Anthropic CEO wants AI companies to slow model development

Anthropic CEO Dario Amodei has called on AI companies to slow the pace of model development, urging more time for strengthening safeguards against misuse. In a recent essay, Amodei argued that companies should ensure safety and alignment measures keep pace with rapidly improving AI capabilities. He proposed a three-step approach, including allowing third-party reviewers inside AI organizations and working together on common safety standards.

A recent Anthropic threat intelligence report detailed instances of Claude AI models being used for activities such as cyberattacks, surveillance, weapons development, and fraud. While the report didn't identify specific actors, it covered cases spanning China, Russia, and Yemen.