OpenAI has disbanded its Preparedness team, which was responsible for assessing the risks posed by the company’s most advanced AI models and developing measures to mitigate them, according to a report from the Financial Times. The reported restructuring was said to have taken place in late July as the company reorganized parts of its safety operation and redistributed responsibility for preparedness work across several specialist teams. OpenAI, however, disputed the central claim in a reply to Engadget, saying the Preparedness team remains in place and that the main change involved its leadership.
Financial Times Report Raises Questions
The Financial Times reported that OpenAI had broken up the dedicated group that evaluated potentially catastrophic risks associated with frontier AI systems, including threats linked to cybersecurity, biological and chemical capabilities, and increasingly autonomous models. Under the reported structure, senior staff in other teams were assigned responsibility for individual preparedness areas rather than concentrating that work within a single standalone unit. The report framed the changes as part of a broader effort to streamline operations as OpenAI focuses resources on its core products and prepares for a possible future public listing.
OpenAI Rejects Disbanding Claim
In a reply to Engadget, OpenAI directly rejected the characterization that the team had been eliminated, saying its preparedness work continues under a centralized safety structure. A company spokesperson told Engadget that OpenAI still has research leaders covering cybersecurity, biological and chemical risks, and AI self-improvement capabilities, with those leaders reporting to Saachi Jain, the company’s head of safety. Engadget subsequently updated its report, stating that Dylan Scandinaro is no longer head of preparedness and that it understood this leadership change to be the only change made to the team.
Preparedness Remains Central to OpenAI’s Safety Framework
The dispute matters because OpenAI’s Preparedness team has historically played a central role in evaluating whether increasingly capable models could create severe or catastrophic harms before deployment. OpenAI’s public safety framework describes preparedness work as covering areas such as biological and chemical threats, cybersecurity risks, and increasingly capable AI systems. The company has also emphasized safeguards, monitoring, evaluations, and mitigation measures as its models become more powerful.
Leadership Departures Intensify Scrutiny
The conflicting accounts arrive amid wider scrutiny of OpenAI’s internal safety organization following several high-profile personnel changes. Ethics lead Chloé Bakalar and safety leader Johannes Heidecke have both left the company, developments that have contributed to questions about how OpenAI is balancing rapid product development, commercial expansion, and long-term risk management. Those concerns have become more prominent as frontier AI systems demonstrate stronger autonomous coding and cybersecurity abilities, increasing pressure on developers to maintain robust testing and governance structures.
For now, the Financial Times reported that OpenAI had disbanded its Preparedness team, while OpenAI rejected that account in a reply to Engadget and said the team continues to operate. The distinction is significant because dissolving the unit would represent a major structural shift in how OpenAI handles catastrophic-risk assessment, whereas replacing its leader would indicate a considerably narrower organizational change. As OpenAI develops more capable systems, the structure and authority of its preparedness function will remain closely watched by researchers, policymakers, employees, and the wider AI safety community.
Source: Engadget