OpenAI Safety Exodus: Head of Safety Systems Departs

Johannes Heidecke, OpenAI's Head of Safety Systems, is leaving the company. This follows a trend of departures in safety leadership at OpenAI. Jan Leike, who led the Superalignment team, left in May 2024, and that team then dissolved. Miles Brundage, who led "AGI readiness," left in October of the same year. Andrea Vallone, a safety research leader, departed a year after Brundage. The "Mission Alignment" team, a successor to Superalignment, disbanded in February 2026. Its former leader, Josh Achiam, resigned four days ago. Wired reports that Mia Glaese, VP and head of alignment, will now lead the teams that previously reported to Heidecke. Saachi Jain will serve as interim head of safety systems under Glaese. OpenAI research chief Mark Chen indicated that increasing demands on safety and faster model training and release cycles are leading to bigger coordination challenges. This constant turnover in safety leadership suggests ongoing organizational shifts within OpenAI.