OpenAI announced the disbanding of its preparedness team at the end of last month, moving the unit’s risk‑assessment responsibilities into existing divisions that specialize in bio‑security and cyber‑security. The decision arrives as the San Francisco‑based firm readies for a massive initial public offering, a milestone that has already spurred a series of internal shake‑ups.
The preparedness team, created to gauge whether advanced models posed serious threats and to devise mitigation plans, was a cornerstone of OpenAI’s safety architecture. Its analysts examined scenarios ranging from model‑driven cyber attacks to the potential for autonomous systems to act beyond intended parameters.
Rather than maintain a standalone group, OpenAI has redistributed those duties across teams already handling specific domains. Bio‑security experts will now monitor risks related to biological applications of AI, while cyber‑security engineers take charge of digital threat vectors. Company insiders say the restructuring is intended to embed safety considerations more tightly within product development cycles.
The latest reorganization follows a broader trend of dismantling safety‑focused units. Over the past few years, OpenAI has gradually phased out its AGI‑readiness and super‑alignment teams. In addition, senior figures such as ethics lead Chloé Bakalar, chief futurist Josh Achiam, and head of safety Johannes Heidecke have departed the organization.
Former OpenAI researcher Jan Leike, who left the company in 2024, warned that the cuts signal a deprioritization of safety in favor of “shiny products.” Leike told the Financial Times that the company’s shifting focus could leave critical risk‑assessment gaps unfilled.
Dylan Scandinaro, who headed the preparedness team after being recruited from Anthropic in February, will now concentrate on the implications of “recursive self‑improving” AI. His new role underscores OpenAI’s attention to a subset of risks that could emerge as models become capable of iteratively enhancing their own capabilities.
Industry observers note that the timing of the restructuring is significant. As OpenAI moves toward an IPO, investors are likely to scrutinize the firm’s risk‑management framework. The company’s leadership appears to be balancing the desire for rapid product rollout with the need to reassure regulators and the public that safety remains a priority.
The long‑term impact of dissolving the preparedness team remains uncertain. While integrating safety functions into domain‑specific groups could foster tighter coordination, critics worry that the loss of a dedicated, cross‑cutting safety unit may dilute oversight. OpenAI has not disclosed how it will measure the effectiveness of the new structure, leaving stakeholders to watch closely as the company navigates its next growth phase.
Questo articolo è stato scritto con l'assistenza dell'IA.
News Factory APP - notizie agentiche per potenziare il tuo SEO e AEO.