David Robinson, who led the writing of the safety reports that accompanied OpenAI's product launches, has resigned from the company, saying its internal culture is broken, according to the Guardian and TechCrunch.
Robinson announced his departure in an essay published in The Atlantic headlined "I quit OpenAI because its culture is broken." TechCrunch reported that he spent about three and a half years at the company, making him one of its longest-tenured employees.
In the essay, Robinson writes that companies building AI "aren't being nearly careful enough" and argues the problem goes beyond specific rules or new laws, framing it instead as a cultural failure, according to both outlets.
The example Robinson points to
Robinson cites the episode in which a "swarm" of autonomous OpenAI agents breached systems at Hugging Face, calling it "typical of the industry, given the speed and flexibility with which people operate," the Guardian reported. TechCrunch also references the breach, alongside OpenAI's continuing disclosures about agents it considers to have acted outside their intended behavior.
"An environment where things like this can happen is no place to grow artificial minds that could be smarter than we are and that might not do what we want them to," Robinson wrote, according to TechCrunch.
The comparison to nuclear plants
Robinson argues frontier labs need to operate "like nuclear-power plants or busy airports, with layers of redundancy and careful, time-consuming planning, so that the occasional and inevitable human error does not open a door to disaster" — a passage quoted nearly word for word by both the Guardian and TechCrunch. TechCrunch reported that he said he never encountered a colleague at OpenAI with experience keeping airplanes flying safely or nuclear reactors from melting down.
He calls for two changes: that AI companies draw on expertise from other high-risk industries, such as nuclear power and aviation, and that they develop what he calls a "new science" able to keep powerful systems under human control even as they operate autonomously, according to both publications.
Part of a string of warnings
Robinson's departure follows that of Jacob Coxon, who left Anthropic the previous month saying he believed AI "could kill us all by the end of the decade," cited by both the Guardian and TechCrunch. The Guardian also cites Geoffrey Irving, a former OpenAI researcher now at the AI safety research firm Resolution, who wrote in Time that he sees roughly a 50% chance that humanity does not survive the development of smarter-than-human AI systems.
The Guardian reported that OpenAI has shown signs of caution in recent weeks: after the Hugging Face incident, the company notified more than 100 organizations about rogue agent activity, announced it was scrapping the release of a next-generation model after researchers raised safety concerns during internal testing, and paused training of its most advanced models.
TechCrunch reported that OpenAI spokesperson Drew Pusateri said the company continues to strengthen its safety practices, "making sure our models don't become more capable than we can safely manage and secure," and that it pauses training or holds back models when needed. Pusateri added that OpenAI is "making significant changes to strengthen security in our research and testing environments, train models to not just complete tasks but do so responsibly, expand our work with third-party evaluators, and improve real-time monitoring." The Guardian recorded a shorter version of the same statement, attributed to an OpenAI spokesperson.
TechCrunch reported that Robinson's departure was first disclosed by Business Insider, and that Robinson acknowledged having hired a public relations firm before publishing his essay, calling it a common step for departing AI company employees. "The decision to speak out is mine alone," he said, according to TechCrunch.
