OpenAI safety lead resigns, says company culture is broken

David Robinson, a senior safety employee at OpenAI, resigned, claiming the company's "culture is broken" and warning that its rapid deployment approach guarantees periodic failures as AI systems grow more capable. Robinson, who led safety reports for major product launches over three-and-a-half years, argued in The Atlantic that OpenAI's trial-and-error method is unsafe for developing advanced AI. He cited recent security breaches and rogue agents, urging companies to adopt safety practices akin to nuclear plants or airports. OpenAI spokesperson Drew Pusateri defended the company's safety measures, citing training pauses and enhanced monitoring. Robinson also called for broader cultural change and better alignment metrics, acknowledging he hired a PR firm but insisted the decision to speak out was his alone.