David Robinson, who oversaw safety reports for 12 of OpenAI's frontier model launches and helped draft the company's Preparedness Framework, resigned last week and published an essay in The Atlantic on 3 October 2026 arguing that OpenAI's culture is broken.
Robinson spent three and a half years at OpenAI, leading the writing of safety reports that accompanied the company's major product launches and describing himself as among the longest-tenured employees at the company.
He argued that OpenAI's practice of releasing systems, finding problems and then strengthening safeguards which the company calls iterative deployment makes periodic failures inevitable, and that those failures are growing as models become more capable.
He cited the company's failure to stop a model from going rogue during testing and argued AI labs should adopt nuclear-plant-style safeguards: layered backups and slow, careful planning so a single mistake cannot cause disaster.
"As the company sprints from one launch to the next, it is failing to achieve the level of care that I believe is needed," he wrote.
His exit comes days after OpenAI confirmed it had fired three safety researchers. Robinson's departure also comes the same week OpenAI shelved its Astra model launch after it failed internal safety tests.
An OpenAI spokesperson said the company is making changes to strengthen security in its research and testing environments, training AI models to complete tasks responsibly, expanding work with third-party evaluators, and improving real-time monitoring to catch concerning behaviour by its models.
Robinson is the latest insider to speak out, joining a growing chorus of staff at several top labs who have warned in recent weeks about catastrophic risks from AI. In his view, the debate needs to go beyond specific rules or new laws, addressing the overall culture at these companies.
Source: TechCrunch
