AI Leaders Back Calls to Pace Frontier Model Development as Safety Concerns Mount

Leaders in advanced AI are converging around a call to slow the pace of frontier model development, arguing that safety systems, monitoring and external oversight are not keeping up with rapidly improving AI agents. Anthropic CEO Dario Amodei laid out the clearest proposal in a September essay, saying the industry should pursue a three-step framework that includes embedded third-party evaluators, coordination among frontier labs in democratic countries and broader global coordination where verification is possible.