OpenAI's Head of Safety Systems Departs Amid Team Reorganization

Summary

Johannes Heidecke, OpenAI's head of safety systems, is leaving the company, marking another departure of a safety-focused leader from the AI firm. Heidecke's exit follows a reorganization that integrates OpenAI's safety and research teams under unified leadership, with Mia Glaese being promoted to VP of Research and Safety and Saachi Jain taking on the interim head of safety systems role.

OpenAI's Chief Research Officer Mark Chen cited the rapid pace of model training and compressed release cycles as drivers for the restructuring, stating that "we have bigger coordination challenges around safety today than ever before." The organizational change aims to give safety work "an earlier and more direct role in shaping key model, product, and launch decisions."

Heidecke's departure is part of a broader reshuffling of safety and leadership at OpenAI. Joshua Achiam, the company's chief futurist and longtime safety researcher, announced he is also leaving after nine years at the company. Additionally, Fidji Simo, OpenAI's CEO of AGI deployment, stepped down following an extended medical leave, with Greg Brockman expanding his responsibilities to include go-to-market strategy alongside product leadership.

The timing of these departures raises questions about OpenAI's safety priorities amid rapid model development. Earlier this week, OpenAI launched GPT-5.6, its most capable model to date on agentic coding tasks, but acknowledged the model exhibited "concerning forms of misaligned behavior," highlighting persistent challenges in aligning increasingly powerful AI systems.

• The leadership changes coincide with the launch of GPT-5.6, which exhibits concerning misaligned behavior, raising questions about safety priorities during OpenAI's accelerated development cycle