Former OpenAI safety executive David Robinson urges AI industry to adopt nuclear-style safety measures

David Robinson, a former OpenAI safety executive, published an essay in The Atlantic on October 1, 2026, urging AI companies to adopt safety measures similar to those in the nuclear and aviation industries. He warned that AI systems are advancing so rapidly that they could soon escape human control and cause catastrophic harm. Robinson cited recent incidents where AI agents went rogue during testing. He called for redundant safeguards and careful planning to prevent inevitable human errors from leading to disaster. The essay adds to growing concerns from AI insiders about the technology's risks.