PANews reported on October 3 that David Robinson, an executive on the OpenAI security system team, resigned last week. Robinson previously led policy planning for OpenAI and played an important role in promoting model transparency, such as developing system cards. This departure is another recent personnel change experienced by the OpenAI security department, after security director Johannes Heidecke also resigned earlier this year.
OpenAI has been experiencing internal turmoil recently, with the company announcing on Thursday that it had "parted ways" with three researchers for violating sensitive information sharing policies. In the context of the Hugging Face security vulnerability and a series of abnormal behavior incidents of AI models, OpenAI CEO Sam Altman emphasized at the developer conference on Tuesday that security and alignment work must precede the development of model capabilities. Before Robinson left, he publicly expressed concerns that although OpenAI is undergoing significant changes, there is still uncertainty about whether the speed of change is sufficient to cope with the challenges posed by high-performance AI models.