MAIN ARTICLE
A departure with consequences beyond one company
David Robinson was no casual observer of frontier AI. According to his own account and reporting by Reuters, The Guardian and TechCrunch, he spent roughly three and a half years at OpenAI, contributed to its preparedness framework and led safety reporting for 12 frontier-model launches. His first-hand experience makes his criticism significant. It does not, however, transform a former employee’s judgment into an independently verified finding of organizational wrongdoing.
In his Atlantic essay, titled “I Quit OpenAI Because Its Culture Is Broken,” Robinson argued that the danger was not merely a technical defect waiting to be patched. He described incentives to move quickly from one launch to another and warned that this culture might diminish attention to early warning signs. His proposed remedy begins with how decisions are made: give safety specialists meaningful authority, plan for rare but consequential failures, and consider the people affected before a system reaches wide distribution.














