West Midlands Police used Microsoft Copilot AI to generate intelligence that included fictitious information about a Maccabi Tel Aviv vs West Ham match that never occurred, which was then presented to decision-makers who banned away fans from attending a real Europa League match.
In October 2025, West Midlands Police prepared intelligence reports for a Safety Advisory Group (SAG) meeting regarding an Aston Villa vs Maccabi Tel Aviv Europa League match scheduled for November 6, 2025. The police used Microsoft Copilot, an AI-powered search tool, which generated false information including a fictitious match between Maccabi Tel Aviv and West Ham United on November 9, 2023 that never took place. This erroneous information was included in written reports presented to the SAG on October 24, 2025, along with other inaccuracies about previous incidents involving Maccabi Tel Aviv fans. The SAG relied on this intelligence to make their decision to ban away fans from attending the match. Initially, Chief Constable Craig Guildford told MPs that the force did not use AI and blamed the error on 'one individual doing one Google search.' However, he later apologized and admitted the mistake 'arose as a result of a use of Microsoft Copilot,' describing it as 'an AI hallucination.' The incident led to a formal inspection by His Majesty's Inspectorate of Constabulary, which found multiple inaccuracies in the police intelligence and criticized the force's handling of the situation. The inspector concluded that confirmation bias influenced the police assessment and that proper intelligence validation procedures were not followed.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
