OpenAI disbands Preparedness team that assessed catastrophic AI risks
The Financial Times reported on August 17 that OpenAI disbanded its Preparedness team, which assessed severe model risks, at the end of July and distributed responsibilities to biosecurity and cybersecurity teams. Lead Dylan Scandinaro will shift to recursive self-improvement research.
On August 17, The Financial Times reported that OpenAI officially disbanded its Preparedness team at the end of last month. The team had been responsible for evaluating whether AI models could cause severe or catastrophic risks and for developing corresponding mitigation measures.
The responsibilities have been divided along specific domains such as biosecurity and cybersecurity and integrated into existing specialist teams. Dylan Scandinaro, who joined OpenAI from Anthropic in February to lead Preparedness, will shift to studying the implications of recursive self-improving AI systems. OpenAI also confirmed it will not retire the Preparedness Framework that governs serious AI risks, though the organizational form responsible for these duties has changed again.
The dissolution is the latest in a series of internal safety restructuring moves. OpenAI previously disbanded Superalignment and other independent safety research units, while Chief Ethicist Chloe Bakalar, Chief Futurist Josh Achiam and safety lead Johannes Heidecke have all departed. OpenAI has now undergone nearly five internal restructurings this year, with several executives who had served less than twelve months leaving in succession. Critics argue that OpenAI is deprioritizing safety as it races toward an IPO in favor of more commercially viable products. OpenAI president Greg Brockman responded on August 17 by saying the company is moving at an extraordinary pace, that every stage requires different leadership, and that he and Altman remain stabilizing forces.