OpenAI reportedly disbanded its preparedness team
| Source: The Verge AI
Tags: OpenAI, AI safety, IPO, preparedness, Jan Leike, Dylan Scandinaro, governance
OpenAI disbanded its dedicated preparedness team in late July — the third major safety unit dissolved in recent years — redistributing risk assessment across bio and cyber teams as it heads toward IPO, with ethics lead Chloé Bakalar, safety head Johannes Heidecke, and Chief Futurist Josh Achiam all having recently departed.
Details
OpenAI shut down its preparedness team at the end of July, according to the Financial Times. The team's mandate was to assess whether models posed serious risks — including rogue behavior and hacking capabilities — and develop mitigation strategies. Responsibility has now been split into domain-specific areas (bio, cyber) and absorbed into existing organizational teams. This is at least the third major safety-focused restructuring at OpenAI in recent years, following the dissolution of the AGI readiness team and the superalignment team. Three senior safety leaders have also departed: ethics lead Chloé Bakalar, Chief Futurist Josh Achiam, and head of safety Johannes Heidecke. Dylan Scandinaro, who was recruited from Anthropic just six months ago to lead preparedness, will now shift focus to the implications of recursive self-improving AI. Jan Leike, who resigned from OpenAI in 2024, told FT that the company is prioritizing 'shiny products' over safety commitments as it heads toward what analysts expect to be a massive IPO. The pattern of dismantling dedicated safety structures — combined with the departure of senior safety personnel — has intensified scrutiny from the AI safety research community about OpenAI's risk management posture at a critical moment.