David Robinson, in an article published in The Atlantic, stated that he worked at OpenAI for three and a half years, making him one of the company's longest-serving employees. He believes that OpenAI's "iterative deployment" approach inevitably leads to periodic failures, and as system capabilities grow, the scale of these failures also expands. Robinson called for frontier AI companies to operate like nuclear power plants or busy airports, with multiple layers of redundancy and meticulous planning, but he noted that he never encountered colleagues with relevant experience at OpenAI. OpenAI spokesperson Drew Pusateri responded that the company is continuously improving its safety measures.