OpenAI disbanded its preparedness team at the end of July 2026, according to a report from the Financial Times. The team was responsible for assessing whether AI models posed serious risks — including the possibility of a model going rogue or hacking another company — and developing ways to mitigate those risks.
Rather than maintaining a dedicated unit, OpenAI has divided the team’s responsibilities across specific domains such as bio and cyber, folding them into existing teams. Dylan Scandinaro, who led the preparedness team after being recruited from Anthropic in February 2026, will now focus on the implications of “recursive self-improving” AI.
The move is the latest in a series of structural changes at OpenAI as the company heads toward what is expected to be a major IPO. In recent years, the company has also dissolved its AGI readiness and superalignment teams. Several safety-focused leaders have departed, including ethics lead Chloé Bakalar, Chief Futurist Josh Achiam, and head of safety Johannes Heidecke.
The departures have drawn criticism from former insiders. Jan Leike, who resigned from OpenAI in 2024, told the Financial Times that the company was ignoring safety in favor of creating “shiny products.”
The dismantling of a team explicitly tasked with evaluating model risks — combined with the exit of multiple safety leaders — may suggest a shift in how OpenAI is prioritizing safety oversight as it scales toward a public offering. Critics could argue the restructuring reduces accountability, though OpenAI has not publicly addressed those concerns in the source reporting.
Source: The Verge