David Robinson, a former safety specialist at OpenAI, has resigned after three and a half years, publicly criticising the company’s approach to managing risk.
In his Atlantic essay ‘I Quit OpenAI Because Its Culture Is Broken’, Robinson argued that the organisation’s fast paced, launch driven culture prioritises speed over care, leaving safety expertise sidelined.
He warned that the “time for trial and error is over,” comparing advanced AI safeguards to those required in nuclear power or aviation. Robinson said OpenAI’s reliance on “iterative deployment” — releasing systems first and fixing problems later — exposes employees and society to avoidable risks.
During his tenure, Robinson helped draft OpenAI’s preparedness framework and oversaw safety reports for 12 frontier model launches. Yet, he concluded that the company was “sprinting from one launch to the next” without achieving the level of care required. His departure adds weight to an industry wide debate: are AI firms moving too quickly, without adequate alignment research or governance?
OpenAI responded that it pauses training or holds back models when safety concerns arise. But Robinson countered that AI capabilities are advancing faster than researchers’ ability to ensure alignment with human goals and values.
For HR leaders, this is more about organisational culture, highlighting the tension between innovation speed and employee confidence in leadership’s duty of care. When safety professionals feel ignored, the risk is not only technical failure but also erosion of trust, disengagement and reputational damage.
Robinson’s exit reminds us that in high stakes industries, culture and governance are as critical as product launches.