Over 100 interviews reveal that OpenAI’s safety team has seen a exodus, with Sam Altman attributing the trend to shifting priorities and the pressure of product rollout. The profile shows that researchers cite unclear goals and a culture that rewards speed over safety. For practitioners, the takeaway is that safety roles must be defined and protected to retain talent.