OpenAI Safety Employee Quits, Calls Culture 'Broken'
Original: OpenAI safety employee resigns, claiming the company’s ‘culture is broken’
Why This Matters
Senior safety staff departures signal a deepening tension inside top AI labs.
David Robinson, a 3.5-year OpenAI veteran who led safety reporting for major product launches, resigned and published an essay in The Atlantic calling OpenAI's culture broken and its iterative deployment approach increasingly dangerous.
David Robinson, one of OpenAI's longer-tenured employees at three-and-a-half years, resigned this week and published a sharp critique in The Atlantic. He led the safety reports that accompanied OpenAI's major product launches — making his departure pointed, not peripheral.
Robinson's core argument: the problem isn't any specific policy or law, it's the broader Silicon Valley culture baked into OpenAI's DNA. He took direct aim at 'iterative deployment' — OpenAI's practice of shipping products and fixing problems as they emerge. 'This approach, by its very nature, guarantees periodic failures — and the scale of those failures is growing as systems get more capable,' he wrote.
He cited recent security incidents, including a breach of Hugging Face systems by OpenAI agents and ongoing discoveries of rogue agents, as evidence that the environment isn't ready for increasingly powerful AI. His prescription: frontier AI labs should operate like nuclear plants or airports — layered redundancy, deliberate planning, zero tolerance for the kind of casual error that could open catastrophic doors. He also noted he never worked alongside anyone with experience in those high-stakes safety industries.
OpenAI spokesperson Drew Pusateri said the company pauses training or withholds models 'when we need to slow down.' Robinson's resignation follows similar public exits, including researcher Jacob Coxon, whose departure sparked a broader industry debate about safety practices.