OpenAI Safety Writer Quits, Urges Nuclear-Level AI Safeguards as Culture Is Called Broken
Updated
Updated · The Verge · Oct 3
OpenAI Safety Writer Quits, Urges Nuclear-Level AI Safeguards as Culture Is Called Broken
3 articles · Updated · The Verge · Oct 3
Summary
David Robinson, who wrote safety reports for every major OpenAI model release, resigned this week and used an Atlantic essay to warn that the company and wider AI industry are mishandling risk.
Robinson said Silicon Valley's "extreme confidence," "perpetual sprints" and "unimpeded optimism" have created a broken culture that underestimates dangers from increasingly capable models.
Nuclear power plants and busy airports are the standard he says frontier labs should emulate, with redundant systems and slow, careful planning so inevitable human errors do not trigger disaster.
His exit follows OpenAI safety-team upheaval reported a day earlier and adds to a broader stream of departures from Anthropic and Google DeepMind as former insiders publicly press for slower, stricter AI development.