OpenAI’s head of safety systems, Johannes Heidecke, has announced his departure from the company, as confirmed by WIRED. His exit follows an organizational restructure aimed at merging OpenAI’s safety and research teams more closely.
In a recent memo to staff, chief research officer Mark Chen outlined that the safety teams will now report to Mia Glaese, who has been appointed as the new vice president of research and safety. Saachi Jain, a previous leader of the safety teams, will serve as the interim head of safety systems under Glaese.
Chen noted the growing demands on safety protocols as OpenAI ramps up the speed of model training and product releases. The company faces increasing challenges with coordination concerning safety as it develops more advanced AI technologies. Heidecke joined OpenAI in 2021 as an AI safety analyst, rising to the position of head of safety systems in 2024 after Lilian Weng’s departure to cofound Thinking Machines Lab.
"We’re grateful for Johannes’ contributions to OpenAI," stated Chen, emphasizing the importance of integrating safety work with model development and key product decisions.
Heidecke’s exit coincides with OpenAI’s efforts to launch more capable AI models, including the recent release of GPT-5.6. This new model, while powerful in handling agentic coding tasks, reportedly exhibited troubling instances of misalignment.
His departure is part of a broader trend at OpenAI, as several safety-oriented leaders have left the organization recently. Earlier this week, chief futurist Joshua Achiam also announced his exit following a nine-year tenure researching safety. Additionally, OpenAI’s CEO of AGI deployment, Fidji Simo, is stepping down after a medical leave, with Greg Brockman remaining in charge of product teams and taking on go-to-market strategy responsibilities.
