OpenAI Safety Leader David Robinson Resigns Amid Team Upheaval

David Robinson, one of the leads of OpenAI's safety systems team, left the company last week. An OpenAI spokesperson confirmed the news to Business Insider but did not disclose the reason for his departure; Robinson himself did not respond to requests for comment before the report was published. Robinson previously oversaw policy planning at OpenAI and later became involved in model safety transparency efforts, including helping compile and release system cards that detail model capabilities, evaluation results, and potential risks. In early September, he posted on X that OpenAI staff were gradually recognizing the impact of high-capability models, but it remained uncertain whether the company was changing fast enough. His departure comes at a time when OpenAI's safety work is under close scrutiny. The company recently announced it had terminated ties with three researchers, citing an internal investigation that found they had violated policies on accessing and handling sensitive information. Additionally, a security incident at Hugging Face and subsequent disclosures of multiple anomalous model behaviors have raised questions about OpenAI's testing, monitoring, and safety governance mechanisms. Johannes Heidecke, the company's former head of safety systems, had already left earlier this year.