OpenAI is reportedly losing the leader of its safety systems team while reorganizing the way safety work connects to model research, a change that comes as scrutiny over AI launches keeps rising.
Johannes Heidecke, who joined OpenAI in 2021, told staff in a memo that he would leave the company, according to the report. His departure comes as OpenAI reshuffles the teams responsible for safety and frontier-model development.
Saachi Jain, who has previously led OpenAI safety teams, is expected to become interim head of safety systems. The report said safety teams will report to Mia Glaese, who is becoming vice president of research and safety.
OpenAI chief research officer Mark Chen said the company wants safety work more closely tied to decisions about models and products. He said it was “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.”
The changes follow OpenAI’s latest model release, GPT-5.6, which was recently cleared by the U.S. government. The company also still lists a Head of Preparedness, a role created to prepare for and reduce severe risks tied to advanced AI systems.
The personnel move lands at a sensitive moment for OpenAI, which is trying to ship new tools quickly while answering questions from regulators, researchers and users about how powerful models are tested before public release.
For users and businesses building around ChatGPT, the reorganization puts a fresh spotlight on who has authority inside OpenAI when safety findings collide with product deadlines.
