OpenAI Reportedly Disbands Its Preparedness Team
OpenAI has dissolved its Preparedness team amid restructuring ahead of a major IPO, following multiple high-profile safety departures.
Stock photo for illustration only, not from the actual event
- OpenAI disbanded its Preparedness safety team at the end of last month.
- Risk assessment duties were divided and moved into existing specialized teams.
- Several key ethics and safety leaders have recently departed the company.
- The reorganization comes as the artificial intelligence firm heads toward an IPO.
OpenAI has reportedly disbanded its preparedness team toward the end of last month, according to information published by the Financial Times. The primary responsibility of the preparedness team was to evaluate whether advanced models presented severe risks and to formulate strategies to mitigate those hazards, such as scenarios where an AI model might go rogue or attempt to hack external companies.
Following the dissolution, responsibilities for specific domains like biological and cyber risks have been segregated and reassigned to pre-existing teams within the organization. This adjustment represents the latest structural shift at the company, which has experienced continuous internal upheaval while preparing for an anticipated large-scale initial public offering.
Stock photo for illustration only, not from the actual event
Over the past few years, the organization has gradually dismantled parts of its research-driven framework, having previously dissolved its AGI readiness and superalignment teams. Furthermore, several prominent figures in safety and ethics have recently exited the firm, including ethics lead Chloé Bakalar, chief futurist Josh Achiam, and head of safety Johannes Heidecke.
Disbanding dedicated safety units during the runway to an initial public offering highlights the commercial pressures AI companies face to prioritize rapid product deployment and profitability over long-term existential risk research, illuminating the tension between safety governance and marketization.
These departures drew criticism from former insiders. Jan Leike, who resigned from the company in 2024, told the Financial Times that the firm was sidelining safety concerns in favor of building flashy products. Meanwhile, Dylan Scandinaro, the head of the disbanded preparedness team who was recruited from Anthropic in February, will transition to focusing on the implications of recursive self-improving artificial intelligence.
Source: The Verge
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment