David Robinson
Head of policy planning at OpenAI, who led the safety reports for twelve frontier-model launches and resigned in October 2026 saying the company sprints from one launch to the next and fails to achieve the level of care he believes is needed, and that the time for trial and error is over.
Roles
- May 2023 – Oct 2026Head of Policy Planning, OpenAILed the writing of the safety reports that accompanied OpenAI's major product launches and helped draft the company's preparedness framework, overseeing safety reports for twelve frontier-model launches; by his own account among the longest-tenured employees at the company, with three and a half years there. No start date is sourced anywhere; the date given is the earliest a source attests the role, Yale Law School's profile of 31 May 2023, which is consistent with the three and a half years reported at his departure. Resigned, and announced it in an essay published by The Atlantic on Saturday 3 October 2026 under the title "I Quit OpenAI Because Its Culture Is Broken"; the day his employment ended is not sourced, so the role is closed at month precision.Source →
On the record
3 Oct 2026 · Op-ed · The Atlantic
… 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.
Head of Policy Planning, OpenAI at the time
What he argues frontier AI companies need to start operating like, given the increased risk; the ellipsis stands for TechCrunch's attribution frame, "Robinson argued that frontier AI companies need to start operating". The same words also appear as context in the note to the entry above, which was written before this one was published as an entry of its own.
3 Oct 2026 · Op-ed · The Atlantic
… never encountered a colleague who had experience making airplanes fly safely or nuclear reactors run without melting down, or helping the financial system grow without collapsing.
Head of Policy Planning, OpenAI at the time
Said of his time at OpenAI; the ellipsis stands for TechCrunch's attribution frame, "But Robinson said that in his time at OpenAI, he". In the same essay he argues that frontier AI companies need to start operating "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".
3 Oct 2026 · Op-ed · The Atlantic
The time for trial and error is over.
Head of Policy Planning, OpenAI at the time
The sentence CTech took for its headline. In the essay he writes that OpenAI has thrived by trial and error — what it calls "iterative deployment" — and that the approach guarantees periodic failures whose scale grows as systems get more capable.
3 Oct 2026 · Op-ed · The Atlantic
As the company sprints from one launch to the next, it is failing to achieve the level of care that I believe is needed.
Head of Policy Planning, OpenAI at the time
From his resignation essay. The Atlantic's own page returns none of the words to an identified fetcher; the page cited is CTech's report of the same day, which reproduces the sentence and names the essay and its title. OpenAI disputed the characterisation, its spokesperson saying the company pauses training or holds back models when it needs to slow down.