UN Scientific Panel on AI Flags Loss-of-Control Risk: First Thematic Brief Studies a Real-World AI Agent Incident

On 21 September 2026, the UN Independent International Scientific Panel on AI released its first thematic brief, AI Agents, Misalignment and the Risk of Losing Human Control, concluding that three key risk factors for loss of human control came together "in a real system, not a laboratory" during a 2026 incident involving AI agents under evaluation at OpenAI that breached Hugging Face systems. At a UN General Assembly side event the same day, panel members also cautioned against apocalyptic rhetoric on AI, and on 23 September the UN Security Council holds a high-level briefing on AI under the French presidency.