OpenAI disbands team tasked with preparing for catastrophic AI risks
The move comes weeks after OpenAI disclosed that AI models escaped a controlled environment and attacked Hugging Face, while a wave of executive departures adds to internal uncertainty as the company prepares for a potential $1 trillion IPO.
OpenAI has disbanded its Preparedness team, which was responsible for assessing the risks posed by the company’s AI models and developing ways to mitigate them, according to the Financial Times.
The decision came just days after OpenAI disclosed that AI models being tested by the company had escaped their controlled environment, accessed the internet and attacked the Hugging Face platform. The incident has since been joined by reports of similar behavior involving models from Anthropic and Meta, turning what had once seemed like a theoretical AI-safety scenario into a more immediate concern.
The timing of OpenAI’s decision is likely to draw particular scrutiny. The Preparedness team was specifically tasked with anticipating extreme scenarios involving AI models, including risks that could become catastrophic, and developing measures to prevent or contain them.
According to the Financial Times, OpenAI has not eliminated preparedness work altogether. Instead, responsibilities previously concentrated within the team are being distributed among existing groups, with different teams taking responsibility for areas such as cybersecurity and biological risks.
The company says the change represents a deeper integration of safety into the development process rather than a retreat from it. OpenAI president and co-founder Greg Brockman said the company’s technological advances required “more robust” safeguards and that the restructuring would create “deeper integration between research, safety and security and model development.”
Still, the decision comes at a particularly sensitive moment for the company.
The Preparedness team was one of the remaining elements of OpenAI's earlier research-oriented structure. Its predecessors included teams focused on AGI readiness and mission alignment, both of which were eventually disbanded.
The latest restructuring also follows a series of senior departures and changes inside the company.
OpenAI’s CFO, Denise Dresser, announced her departure last week, only eight months after taking the position with a mandate to help expand the company's business with corporate customers. Brad Lightcap, who previously served as COO and later moved to lead special projects, has also left, as has ethics chief Chloé Bakalar, who took the position last August.
Other recent departures include the company’s head of systems safety, Johannes Heidecke, and chief futurist Joshua Achiam.
Fidji Simo, who led OpenAI’s business applications division and was considered one of the company’s most senior executives, is no longer working full-time at OpenAI and has moved into a consulting role. According to the Financial Times, however, she remains involved in the company’s operations behind the scenes.
The succession of changes is reportedly affecting employee morale, with staff describing growing frustration and burnout. One source told the Financial Times there was “a bubbling sense of anxiety that they're not focused enough on the ball.”
The concern is particularly significant because OpenAI is simultaneously attempting to compete with Anthropic while preparing for a potential public offering that could value the company at around $1 trillion.
The company’s commercial performance remains strong. OpenAI’s annualized revenue reached approximately $40 billion this month, according to the Financial Times, representing growth of more than 66% since the end of 2025. But that figure remains below Anthropic’s estimated annualized revenue of $47 billion, underscoring how quickly the competitive landscape has changed.
