OpenAI’s safety chief is leaving, the 6th high-profile exit this year

OpenAI’s executive suite is experiencing another setback. According to a report by Wired, Johannes Heidecke, the company’s head of safety systems, has told staff he is leaving, just after an internal reorganisation that merged OpenAI’s safety and research divisions.

In an internal memo, OpenAI’s Chief Research Officer, Mark Chen, said the company’s safety teams would now be consolidated under Mia Glaese, the head of alignment, who steps into an expanded role as VP of Research and Safety. Saachi Jain will take over as interim head of safety systems. ‘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. As a result, we have bigger coordination challenges around safety today than ever before,’ Chen said.

Heidecke’s exit ends a five-year stint at OpenAI that began in 2021. He stepped into the safety chief role in 2024 after his predecessor, Lilian Weng, left to co-found rival startup Thinking Machines Lab.

His departure follows that of Joshua Achiam, OpenAI’s Chief Futurist, who announced days earlier he was leaving after nine years of safety research, and Fidji Simo, CEO of AGI Deployment, who is stepping down following an extended medical leave. Co-founder Greg Brockman will absorb her responsibilities, taking control of product teams and OpenAI’s global go-to-market strategy.

The shakeup underscores the mounting operational pressure inside OpenAI as it races to build and deploy advanced AI models while eyeing a highly anticipated IPO.

Image: Wikimedia Commons/by Loic Le Meur