OpenAI safety veteran quits, says the ‘time for trial and error is over’
David Robinson, a former OpenAI safety employee who helped draft the company's preparedness framework, resigned and criticised its fast-moving culture in an Atlantic essay published Oct 3, arguing AI firms are not being “nearly careful enough.”
Robinson, who spent three and a half years at the company overseeing safety reports for 12 frontier-model launches, said OpenAI relies heavily on “iterative deployment” — releasing systems and tightening safeguards only when problems emerge. He argued that approach no longer matches the risks.
He wrote that advanced AI systems need safeguards closer to those in nuclear power and aviation, and that “the time for trial and error is over.” An OpenAI spokesperson responded that the company pauses training or holds back models when needed.
The resignation adds fuel to an industry-wide debate over AI safety, following scrutiny of OpenAI and Anthropic after safety incidents. It comes alongside the White House's new “Super Intelligence Force” task force examining AI's risks and opportunities.
