OpenAI dissolved the team built to catch catastrophic AI risks, reassigning its work to other groups

What happened

OpenAI has dissolved its internal “Preparedness” team, which focused on identifying and evaluating catastrophic risks posed by its own AI models. The responsibilities of this specialized group are now distributed among other teams within the company. Following this move, several safety-focused staff members have opted to leave OpenAI, signaling internal unease about the shift in approach to AI safety.

Why it matters

This change weakens OpenAI’s dedicated capacity to catch and manage extreme AI risks. The Preparedness team was designed to serve as an early warning system for threats that might arise from highly capable AI systems. By reallocating that work across teams with broader priorities, OpenAI risks diluting focus and slowing critical risk detection. For operators, investors, and regulators, this move raises the cost of trusting OpenAI’s safety commitments and puts pressure on external actors to increase scrutiny and safety oversight.

What to watch next

Monitor whether the newly reassigned groups maintain rigorous threat modeling and mitigation efforts at the level the Preparedness team demanded. Watch for further departures of safety personnel, which could further erode in-house expertise. Investors and enterprise customers should watch for shifts in OpenAI’s risk disclosures and governance updates. Regulators might also see this as a signal to tighten AI safety standards or demand clearer accountability on catastrophic risk management from major AI developers.

AI Quick Briefs Editorial Desk