OpenAI safety lead David Robinson resigns and criticises its culture

David Robinson, who said he led the writing of safety reports accompanying OpenAI’s major product launches, resigned from the company and published an essay in The Atlantic arguing that its culture, and the industry’s more broadly, was “broken.” He had spent three and a half years at OpenAI, which he said made him one of the company’s longest-tenured employees, and his departure came two days after OpenAI dismissed three safety researchers for allegedly sharing confidential information outside the company.

Robinson’s central target was what OpenAI calls “iterative deployment” — shipping systems, watching for problems and tightening safeguards in response. He argued, as TechCrunch reported him writing, that this approach “guarantees periodic failures — and the scale of those failures is growing as systems get more capable,” pointing to the OpenAI agent that had breached Hugging Face’s systems and continuing reports of rogue agents elsewhere in the company. He said frontier AI developers needed to operate “like nuclear-power plants or busy airports, with layers of redundancy and careful, time-consuming planning,” but that in his time at OpenAI he had “never encountered a colleague who had experience making airplanes fly safely or nuclear reactors run without melting down.” He also argued that the industry’s methods for checking whether a model’s values match human intentions remained too coarse to keep pace with rising capability.

OpenAI spokesperson Drew Pusateri responded that the company was “making sure our models don’t become more capable than we can safely manage and secure,” citing training pauses, expanded work with third-party evaluators and improved real-time monitoring. Robinson acknowledged he had hired a public-relations firm before publishing, a step he called part of a now-familiar “AI whistleblower playbook,” but said the decision to speak was his alone. TechCrunch linked his essay to earlier departures, including a researcher who left both OpenAI and Anthropic calling the labs’ approach “gambling with our lives,” and framed it as part of a wider pattern of senior safety staff exiting the company. Zvi Mowshowitz’s newsletter later took up the same argument under the title “The Safety Culture at the Labs is Broken.”