tech
The leaders responsible for keeping OpenAI's AI safe keep leaving
OpenAI just lost another leader of one of its safety teams. The company's safety and alignment teams have had high turnover in recent years.
TL;DR
- Johannes Heidecke, head of OpenAI's Safety Systems team, is leaving the company.
- OpenAI is reorganizing its safety and research work under Mia Glaese, its new VP of research and safety.
- This move integrates safety more deeply into research teams.
- The company aims to ensure safety decisions are informed by model capabilities and vice versa.
- OpenAI has a notable track record of high turnover among its safety leaders.
- Previous departures, like Jan Leike and Ilya Sutskever, included public criticisms of OpenAI prioritizing products over safety.
- Other notable departures from OpenAI's safety teams include Brundage, Weng, Vallone, Madry, and Achiam.
- Some former employees have joined competitor Anthropic, which also experiences departures but maintains core safety leadership more consistently.
- The article lists eight leaders of OpenAI's safety-focused teams who have left in recent years.