This week, Johannes Heidecke, OpenAI's head of safety systems, announced his departure from the organization, as reported by WIRED. His exit follows a restructuring initiative aimed at combining the safety and research departments within the company.
In a memo distributed to employees and obtained by WIRED, Mark Chen, the chief research officer, indicated that the safety teams will now be overseen by Mia Glaese, who serves as VP of research and head of alignment. Glaese will assume an expanded position that encompasses both research and safety responsibilities. Saachi Jain, who previously managed the safety teams, has been appointed as the interim head of safety systems and will report directly to Glaese.
Chen emphasized the increasing demands on safety, noting, “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.”
Heidecke, who originally joined OpenAI in 2021 as an AI safety analyst, took over leadership of the safety systems in 2024 after his predecessor, Lilian Weng, departed to co-found Thinking Machines Lab with fellow OpenAI researchers.
“We appreciate Johannes’ significant contributions to OpenAI,” Chen stated. “It’s crucial for our safety initiatives to be integrated with the development of advanced models, ensuring a more direct role in critical decisions regarding models, products, and launches. We look forward to this new phase under Mia Glaese’s leadership across both research and safety.”
As part of this transition, OpenAI is aiming to release increasingly sophisticated AI models. Just recently, the company unveiled GPT-5.6, its most advanced model yet for agentic coding tasks, although it exhibited some troubling instances of misalignment compared to earlier models.
Heidecke is not the only leader focused on safety to leave OpenAI recently; earlier in the week, Joshua Achiam, the chief futurist, notified his colleagues of his intention to exit after nine years of work in safety research.
Moreover, changes are occurring in other areas of the company as well. Fidji Simo, the CEO of AGI deployment, announced her resignation after an extended medical leave. In her absence, Greg Brockman will continue to manage OpenAI's product teams while also taking on responsibilities related to go-to-market strategy.
