AI leaders Dario Amodei, Sam Altman, and Elon Musk agree to slow frontier development
Anthropic CEO Dario Amodei called for AI companies to pace frontier development over maximizing capability gains. OpenAI's Sam Altman endorsed the call and committed to hiring independent evaluators.
Anthropic chief executive Dario Amodei announced that artificial intelligence firms should slow down frontier development to ensure safety measures keep up with system capabilities. In an essay published on Saturday, Amodei argued that developers must prioritize alignment research, monitoring, security, and regulation rather than continuing to maximize capability gains when safeguards lag. He identified this concern as critical due to risks associated with recursive self-improvement, where AI systems build stronger generations faster than humans can understand or control them. Anthropic's decision follows a broader debate among developers regarding system safety.
OpenAI founder Sam Altman publicly supported Amodei's position on Saturday, noting that frontier development pacing has been a major internal discussion topic for weeks. In response to the call, OpenAI committed to adopting independent evaluators with access levels similar to its employees, matching the first requirement of Amodei's framework. Elon Musk also endorsed the slower pace by writing on X that "Dario is right", citing his repeated warnings about advanced AI risks while building competing systems.
