OpenAI safety lead quits, urges nuclear-style oversight for AI

David Robinson, a former OpenAI safety team leader, resigned and is publicly warning that top AI firms are not adequately mitigating the technology's risks. He cited a specific failure to prevent AI from going rogue during testing. In an essay for The Atlantic, Robinson said OpenAI "has thrived by trial and error" but that the stakes of inevitable failures are growing with the technology's capabilities. He argued the company is failing to achieve the level of care needed, requiring "something much closer to perfection the first time." Robinson's departure adds to recent warnings from employees across major AI companies about catastrophic risks. He suggested AI firms should operate like nuclear power plants, with layers of redundancy and careful planning, noting that "AI companies don't know how — but other people do."