As OpenAI races toward an anticipated high-stakes initial public offering, the artificial intelligence powerhouse is once again restructuring its internal safeguards, raising fresh concerns among industry observers about the company's commitment to risk management. According to a report by the Financial Times, OpenAI dismantled its dedicated preparedness team late last month, distributing its oversight duties across existing divisions.
The preparedness team was tasked with evaluating catastrophic risks associated with frontier AI models, including potential scenarios where autonomous systems might breach cybersecurity defenses or assist in bioweapon development. Under the new operational structure, responsibility for specific threat vectors—such as bio and cyber risks—has been reassigned to existing specialized teams within the broader organization. Meanwhile, Dylan Scandinaro, the head of the preparedness team who was hired from rival Anthropic in February, will pivot to analyzing the implications of recursive self-improving artificial intelligence.
This restructuring marks the latest in a series of organizational shifts within OpenAI's safety apparatus. Over the past few years, the company has gradually dissolved key safety initiatives, including its AGI readiness and superalignment teams. The reorganization coincides with a recent departure of several safety leaders, including ethics lead Chloé Bakalar, Chief Futurist Josh Achiam, and head of safety Johannes Heidecke.
The move has drawn sharp criticism from former safety personnel and industry watchdogs. Jan Leike, a prominent researcher who resigned from OpenAI in 2024, told the Financial Times that the company appears to be sidelining safety culture in favor of deploying "shiny products." As AI systems rapidly advance, OpenAI's organizational shift underscores the growing tension between rapid commercialization and rigorous risk oversight in the tech industry.