According to a report by the Financial Times on August 17, OpenAI officially disbanded its Prevention Team at the end of last month. The team was originally responsible for assessing whether frontier models posed serious risks such as bioweapons or cyberattacks, and developing corresponding mitigation strategies. After the disbandment, its functions were split according to risk areas such as biological and cyber threats, and integrated into existing business teams.

OpenAI

This is the latest move in OpenAI's recent organizational restructuring. Over the past few years, the company has successively disbanded the AGI Preparedness Team and the Super Alignment Team, gradually moving away from its early public governance structure that emphasized safety. At the same time, key personnel in critical safety positions such as Ethics Lead Chloé Bakalar, Chief Futurist Josh Achiam, and Head of Security Johannes Heidecke have recently left the company. Jan Leike, the former head of the Super Alignment Team who left in 2024, told the Financial Times that the company is systematically neglecting safety investments in pursuit of "cool products."

After the disbandment of the Prevention Team, Dylan Scandinaro, the newly appointed head of the Preparedness Team who joined OpenAI from Anthropic in February, will focus on researching the potential impacts of "recursive self-improving" AI. These series of changes have occurred at a crucial moment when OpenAI is preparing for a large-scale IPO, and the company's governance structure is accelerating its transition from a non-profit-led model to a commercial entity, with the tension between safety and growth remaining a focal point of external attention.