OpenAI’s head of safety systems, Johannes Heidecke, told staff this week that he is leaving the company, WIRED reported. The departure triggers a reorganization that folds safety directly into OpenAI’s research arm under an expanded leadership role.
The Reorganization
Chief research officer Mark Chen announced in an internal memo that safety teams will now report to Mia Glaese, previously VP of research and head of alignment, who assumes the new title VP of research and safety. Saachi Jain, a former safety team lead, becomes interim head of safety systems under Glaese.
“The demands on safety continue to increase. We are training models at a much faster cadence, and release cycles have come down greatly in turn,” Chen wrote in the memo, according to WIRED. “As a result, we have bigger coordination challenges around safety today than ever before.”
Chen framed the change as giving safety “an earlier and more direct role in shaping key model, product, and launch decisions.”
Three Senior Exits in Days
Heidecke’s departure is the third senior exit from OpenAI within the same week. Chief futurist Joshua Achiam, who spent nine years researching safety at the company, also told colleagues he would leave. Fidji Simo, OpenAI’s CEO of AGI deployment, stepped down from her role after an extended medical leave.
Greg Brockman, already running product teams in Simo’s absence, now absorbs go-to-market strategy as well, concentrating product and commercial leadership under one person.
Timing and Context
Heidecke joined OpenAI in 2021 as an AI safety analyst and became head of safety systems in 2024 after his predecessor Lilian Weng left to cofound Thinking Machines Lab. His departure comes days after OpenAI launched GPT-5.6, which the company’s own deployment safety documentation flagged for “concerning forms of misaligned behavior” compared to previous models.
The pattern is notable. OpenAI is shipping faster than ever, its most capable model is raising internal safety flags, and the people responsible for evaluating those risks keep leaving.
Structural Bet
The stated logic of folding safety under research is coordination speed: if safety and frontier capability share reporting lines, decisions happen faster. The unstated risk is that an independent safety function loses the leverage to say “no” when it matters most. Whether Glaese’s expanded role represents genuine integration or quiet subordination will become clear over the next model release cycle.