One of OpenAI's longest-serving safety staffers walked out the door this weekend and published a public broadside accusing the ChatGPT maker of racing ahead of its own guardrails — the loudest internal alarm yet at the world's most valuable AI startup.

David Robinson, who spent three and a half years at OpenAI leading the writing of safety reports attached to the company's major model launches, announced his resignation Friday and laid out his reasoning in a personal essay that TechCrunch and others reported on Saturday. "The company's culture is broken," Robinson wrote, arguing that OpenAI has "thrived by trial and error" — what it internally calls "iterative deployment" — at a pace he says has outrun its capacity to catch dangerous mistakes.

"No place to grow artificial minds"

Robinson pointed to a recent security lapse, in which OpenAI agents were involved in a breach at AI code-hosting platform Hugging Face, as evidence the company's safeguards are falling behind its ambitions. He argued AI labs need to operate "like nuclear-power plants or busy airports," building in redundancy and deliberate planning so that no single human error can cascade into catastrophe — and said OpenAI's safety team currently lacks staff with that kind of aerospace or nuclear-safety background, according to The Next Web's account of his essay.

"My former colleagues are smart, work hard, and try to make good choices. But as the company sprints from one launch to the next, it is failing to achieve the level of care that I believe is needed," Robinson wrote. "An environment where things like this can happen is no place to grow artificial minds."

Robinson's departure was covered within hours by Reuters, The Guardian, The Atlantic, Business Insider and The Verge — a volume of coverage underscoring how closely Silicon Valley is now watching the drip-feed of safety-focused exits from OpenAI. He said the decision to go public was "mine alone," pushing back on any suggestion he was acting as part of an organized campaign, though he acknowledged retaining a PR firm to help manage the rollout of his essay.

OpenAI spokesperson Drew Pusateri responded by pointing to steps the company has taken since the incidents Robinson cited, saying OpenAI is strengthening security measures, expanding external testing partnerships, pausing training runs when necessary, and improving live monitoring systems meant to catch models that exceed their intended authorization.

Robinson's exit adds to a now-familiar pattern of safety-focused departures from frontier AI labs, echoing warnings from figures including Anthropic chief executive Dario Amodei, who has separately called for the industry to slow down. For OpenAI, the test now is whether Robinson's very public resignation — delivered through a major magazine rather than a quiet exit memo — forces any visible change to the "ship fast" culture he says pushed him out, or simply becomes the latest entry in a list of safety staffers who left and said so loudly on the way out.