OpenAI is restructuring its artificial intelligence safety division, consolidating research and safety operations under a single leadership framework. The reorganization follows the departure of Johannes Heidecke, head of the company’s Safety Systems team, continuing a pattern of high-profile exits among OpenAI’s alignment personnel. Chief Research Officer Mark Chen confirmed that Heidecke’s transition supports the company’s initiative to embed safety protocols directly into broader research workflows. Mia Glaese has been appointed vice president of research and safety to oversee the unified division. Company representatives stated the restructuring reflects a strategic belief that safety evaluation and model capability development must be managed by integrated leadership to ensure informed decision-making. OpenAI maintains that mitigating advanced AI risks remains central to its mission of ensuring artificial general intelligence benefits humanity. The leadership changes arrive amid sustained criticism that commercial deployment timelines have increasingly overshadowed long-term safety planning. Heidecke’s exit adds to a notable roster of former safety experts who have left the organization or publicly questioned its priorities. In recent years, cofounders Ilya Sutskever and Jan Leike departed to launch independent alignment initiatives, with Leike explicitly citing a cultural shift toward product releases over rigorous risk preparation. Senior advisor Miles Brundage and former vice president of research and safety Lilian Weng also left, with Brundage warning that global institutions remain unprepared for frontier AI capabilities. Additional departures include Andrea Vallone, who moved to Anthropic, former chief futurist Joshua Achiam, who stated he will pursue alignment work externally, and Aleksander Madry, who was previously reassigned from his preparedness role before leaving the company. This leadership churn has extended beyond OpenAI, reflecting broader industry tensions. Anthropic, established by former OpenAI executive Dario Amodei following safety disagreements, has also faced notable exits, including safeguards lead Mrinank Sharm, who cautioned against industry-wide pressure to deprioritize alignment. Despite these widespread challenges, industry analysts observe that Anthropic has maintained comparatively stable safety leadership. OpenAI’s consolidation under Glaese represents a structural response to internal and external scrutiny, aiming to streamline accountability while accelerating model development. Nevertheless, the continued departure of specialized safety personnel highlights enduring friction between rapid commercialization and comprehensive risk governance. As frontier AI systems grow increasingly capable, the industry remains focused on whether corporate restructuring can effectively balance innovation velocity with the rigorous oversight required for safe technological advancement.