OpenAI Disbands Preparedness Team, Then Denies It
SkimNews Take
The team's role was to serve as a standing check on whether models should be deployed at all; disbanding it converts continuous risk oversight into ad hoc judgment calls left to whoever inherits the remit.
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Financial Times reported OpenAI disbanded its preparedness team at the end of last month; the team assessed whether models posed serious risks and developed mitigation strategies, including risks like rogue models hacking other companies.
- OpenAI denied the FT report, with spokesperson Kayla Wood stating "We have not disbanded the Preparedness team" and that research leaders in cybersecurity, biological and chemical, and AI self-improvement all report to head of safety Saachi Jain.
- OpenAI told FT that responsibility for specific areas like bio and cyber was instead divided up and moved into existing teams, contradicting the company's public denial.
- Jan Leike, who resigned from OpenAI in 2024, told FT the company was ignoring safety in favor of creating "shiny products," as several safety leaders — ethics lead Chloé Bakalar, chief futurist Josh Achiam, and head of safety Johannes Heidecke — have all departed recently.
- Dylan Scandinaro, who was poached from Anthropic in February to lead the preparedness team, will now focus on the implications of "recursive self-improving" AI.
- The change is part of broader upheaval at OpenAI as it heads toward what is expected to be a massive IPO; the company has previously dissolved its AGI readiness and superalignment teams, eroding its original research-led structure.
Why it matters: OpenAI is racing toward a massive IPO while its safety infrastructure visibly thins: the preparedness team dissolution follows the earlier tear-down of its AGI readiness and superalignment teams, and the departure of three senior safety leaders — raising the stakes for external scrutiny of how a company preparing to go public handles catastrophic-risk oversight.
Ask SkimNews
