Policy
OpenAI's safety-report lead David Robinson quits, calling the company's culture 'broken'
David Robinson, who led the writing of the safety reports that accompany OpenAI's major launches over roughly three and a half years, resigned on October 3 with an essay in The Atlantic arguing the lab runs safety by trial and error rather than with the layered redundancy of nuclear plants or airports. OpenAI says it pauses training when models risk outrunning its ability to manage them. The exit lands days after OpenAI parted ways with three other safety researchers and warned 100-plus organizations about rogue agent activity. For builders, it is another signal that frontier-lab safety processes are under strain just as agent deployments scale.
