OpenAI safety report author David Robinson leaves and publicly criticizes industry culture

David Robinson, who was responsible at OpenAI for writing safety reports for every major model release, resigned this week and published an opinion piece in The Atlantic saying the industry's culture has fundamentally collapsed. He believes the problem goes deeper than adding a few training rules or regulations, and that Silicon Valley keeps scaling up large models with extreme confidence and continuous sprinting, under unconstrained optimism, ignoring or underestimating potential problems. He argues that frontier labs should operate like nuclear power plants or busy airports, with multiple layers of redundancy and time-consuming careful planning, so that inevitable human errors do not open the door to disaster. The article also mentioned that researchers and safety staff including Anthropic's Jacob Coxon, Joe Benton and Google DeepMind's Robert O'Callahan, Bilal Chughtai, Josh Engels had previously left and spoken out publicly.