David Robinson, a leader on OpenAI's Safety Systems team, resigned last week. An OpenAI spokesperson confirmed the departure on Friday.
Robinson's exit marks another departure from the company's safety ranks. He had focused on safety transparency, developing and sharing system cards that document OpenAI's AI model capabilities and limitations.
The departure comes as OpenAI navigates multiple safety pressures. On Thursday, the company said it had "parted ways" with three researchers for violating policies on sharing sensitive information. Earlier departures include Johannes Heidecke, who headed safety efforts, and the May 2024 exits of co-founder Ilya Sutskever and researcher Jan Leike, whose departure led to the dissolution of their superalignment team.
Robinson himself signaled discontent in early September, writing on X that people at OpenAI were "waking up" to the implications of highly capable AI models. "I do not know whether we are changing fast enough," he wrote.
CEO Sam Altman addressed the tension directly at the company's developer conference on Tuesday, saying safety and alignment must "continue to stay ahead of model capabilities." That formulation—staying ahead—acknowledges the gap Robinson and others saw.
The pattern of exits from critical safety and policy teams suggests structural misalignment between the pace of model capability development and the company's capacity to govern it. For a company whose primary competitive advantage rests on customer trust and regulatory forbearance, repeated safety departures create real business risk.
