Former OpenAI safety employee warns of a broken industry culture
David Robinson, who previously prepared safety reports for every major OpenAI model release, resigned from the company this week. He is now discussing his concerns in an editorial published by The Atlantic, arguing that the culture surrounding AI development is fundamentally broken.
Robinson says the problem goes beyond adding new rules or regulations for training models. In his view, the technology industry’s confidence, rapid development cycles, and optimism have encouraged companies to overlook or underestimate possible dangers. He calls for greater humility and for AI developers to learn from industries that manage high-consequence risks.
Robinson argues that frontier AI laboratories should use safeguards comparable to those at nuclear power plants or busy airports. He says multiple layers of redundancy and deliberate planning are needed so that an ordinary human error does not lead to a disaster. His departure follows a series of resignations by safety researchers from major AI companies, including Jacob Coxon, Robert O’Callahan, Bilal Chughtai, Josh Engels, and Joe Benton, who have also made public warnings about AI risks.
Written by Sakhanda Wire AI from reporting by The Verge. Facts, quotes and full context are in the original article below.