OpenAI has quietly dismantled its preparedness team, the group tasked with assessing whether its models posed serious risks and developing ways to mitigate them. The Financial Times reports the team was disbanded at the end of last month, with its responsibilities carved up by domain — biological threats, cybersecurity, and so on — and folded into existing teams. It’s the latest in a pattern that critics say has been building for years.
The moves come as OpenAI barrels toward what’s expected to be a massive IPO, and the timing hasn’t gone unnoticed. The company previously dissolved its AGI readiness and superalignment teams, and a string of senior safety figures have walked out the door. Ethics lead Chloé Bakalar, Chief Futurist Josh Achiam, and head of safety Johannes Heidecke have all departed recently. Jan Leike, who resigned from OpenAI in 2024, told the FT the company was ignoring safety in favor of creating “shiny products.”
Dylan Scandinaro, who led the preparedness team after being poached from Anthropic in February, will now focus on the implications of “recursive self-improving” AI. That’s a telling assignment — recursive self-improvement is the scenario where AI systems get better at improving themselves, which has long been a core concern among researchers worried about losing control. OpenAI is keeping an eye on that particular risk, but the organizational muscle to do anything about it looks thinner than it did a month ago.
Some of this is standard corporate restructuring, and OpenAI isn’t the first company to fold specialized teams into broader groups ahead of going public. But the pattern is hard to ignore: safety teams keep shrinking while product launches keep accelerating. The question that hangs over the IPO isn’t whether OpenAI can make money — it’s whether the company that promised to build AI safely still has the internal structure to keep that promise. Leike’s “shiny products” jab stings precisely because it’s the kind of criticism that used to come from outside the company, not from the people who once ran its safety efforts.
