OpenAI disclosed that Johannes Heidecke, the organization’s head of safety systems, will exit his role as part of a sweeping restructuring of its safety and research divisions. The announcement came via a staff memo that Heidecke shared with employees, a document seen by Wired. Heidecke, who entered OpenAI in 2021, has overseen safety initiatives through the launch of several flagship models.
Following his departure, Saachi Jain will assume the position of interim head of safety systems. Jain, who previously led OpenAI’s safety teams, is expected to maintain continuity while the company finalizes its new leadership framework.
In the revised hierarchy, all safety groups will report to Mia Glaese, who will serve as the new vice president of research and safety. Glaese’s appointment consolidates research and safety under a single executive, a move the company says will streamline decision‑making on model development, product design, and launch strategies.
Mark Chen, OpenAI’s chief research officer, emphasized the rationale behind the shift. "It’s important that our safety work is integrated with frontier‑model development, with an earlier and more direct role in shaping key model, product and launch decisions," Chen told Wired. The statement signals a strategic pivot toward embedding safety considerations at the earliest stages of AI creation.
The reorganization arrives shortly after OpenAI released GPT‑5.6, its latest language model, which recently received approval from the U.S. government. The timing suggests the company aims to align its safety infrastructure with the heightened scrutiny and regulatory expectations surrounding advanced AI systems.
OpenAI retains a Head of Preparedness, a role filled earlier this year to “prepare for and mitigate … severe risks,” according to CEO Sam Altman’s post on X. The position underscores the firm’s ongoing commitment to risk assessment even as it reshapes its safety leadership.
Company officials indicated the changes are designed to foster faster, more coordinated responses to emerging safety challenges while preserving the innovative pace that defines OpenAI’s research agenda. No timeline for a permanent replacement of Heidecke’s role was provided, and the organization said it will continue to evaluate its structure as the AI landscape evolves.
Cet article a été rédigé avec l'assistance de l'IA.
News Factory APP - actualités agentiques pour booster votre SEO et AEO.