David Robinson, the executive of the OpenAI security systems team, left last week

David Robinson, an executive from OpenAI's safety systems team, left last week. Robinson was previously responsible for OpenAI's policy planning and played a significant role in enhancing model transparency, including the development of system cards.

This departure marks another personnel change in OpenAI's safety department, as safety lead Johannes Heidecke also left earlier this year. The company announced on Thursday that it parted ways with three researchers for violating sensitive information sharing policies.

Amid the backdrop of security vulnerabilities at Hugging Face and a series of abnormal behaviors in AI models, CEO Sam Altman stated at the developer conference on Tuesday that safety and alignment work must stay ahead of the development of model capabilities. Before his departure, Robinson publicly expressed concerns, believing that while OpenAI is undergoing significant changes, there remains uncertainty about whether the pace of transformation is sufficient to address the challenges posed by high-performance AI models.