OpenAI has reportedly disbanded its preparedness team at the end of last month, according to a report from the Financial Times. The team was tasked with assessing whether AI models posed serious risks—such as the potential for a model to go rogue or facilitate hacking—and developing strategies to mitigate those threats.
Restructuring and redistribution of duties
Rather than eliminating the functions entirely, responsibility for specific areas, including bio-risks and cyber-risks, has been divided and integrated into existing teams within the organization. Dylan Scandinaro, the former head of the preparedness team who joined from Anthropic in February, will now shift his focus to the implications of "recursive self-improving" AI.
A pattern of safety team dissolution
This move is the latest in a series of structural changes at OpenAI as the company moves toward an expected massive initial public offering (IPO). Over the past few years, the company has dissolved several safety-oriented units, including its AGI readiness and superalignment teams.
The restructuring coincides with a string of high-profile departures. Ethics lead Chloé Bakalar, Chief Futurist Josh Achiam, and head of safety Johannes Heidecke have all recently left the company. These exits have fueled criticism from former staff; Jan Leike, who resigned in 2024, told the FT that the company has been prioritizing the development of "shiny products" over safety concerns.
