David Robinson, who led safety reports for major OpenAI product launches, is leaving after three and a half years, saying the company’s culture is broken. He argues that addressing AI risks requires changing corporate culture, not merely adopting specific rules or new laws.