OpenAI safety employee resigns, claiming the company’s ‘culture is broken’
techcrunch ·
David Robinson, a long-tenured safety employee at OpenAI, has resigned and published an essay in The Atlantic claiming the company's culture is broken. He argues that OpenAI’s reliance on iterative deployment guarantees periodic failures as AI systems become more capable, citing recent security breaches involving rogue agents. Robinson contends that frontier AI companies should operate with the rigorous redundancy found in nuclear power plants or airports, rather than relying on trial and error. He criticizes the industry for lacking colleagues with experience in high-stakes safety-critical fields like aviation or nuclear energy. While OpenAI spokesperson Drew Pusateri responded by highlighting ongoing improvements to safety measures and real-time monitoring, Robinson insists that current alignment metrics are too coarse. He warns that allowing models to grow smarter while these fundamental issues remain unsolved increases danger. Robinson also noted that he hired a PR firm but maintained that his decision to speak out was entirely his own.
techcrunch