OpenAI's head of safety, Johannes Heidecke, is leaving the company as it folds its research and safety teams into a single organization, according to a report by Maxwell Zeff in Wired.
As part of the shake-up, Mia Glaese will step into a newly framed role as vice president of research and safety, per Wired. The reporting frames Heidecke's exit as coming while OpenAI works to more tightly integrate the two groups that had previously operated on separate tracks.
Bloomberg, citing Wired, described the move as a departure following a reshuffle at the company. Beyond the personnel change and the reorganization itself, the sources do not detail Heidecke's reasons for leaving or where he is headed next.
Why does a management reshuffle at one company warrant attention? Because OpenAI, the maker of ChatGPT, sits at the center of the debate over how fast and how safely powerful AI systems should be built. How a leading lab structures its safety function — whether it stands apart from research or is woven directly into it — signals how the company thinks about balancing rapid development against caution.
Supporters of merging the teams often argue that safety work is more effective when it is embedded alongside the researchers building the models, rather than siloed. Critics of such moves tend to worry that combining the functions can dilute the independence of safety oversight. The sources here do not state which interpretation applies to OpenAI's decision.
The departure of a safety lead, combined with a restructuring of who owns safety, matters because it offers a rare window into how one of the most influential AI companies is choosing to organize the people responsible for keeping its technology in check.