West Midlands Police use fabricated AI output to ban Israeli fans from football match
West Midlands Police recommended banning Maccabi Tel Aviv fans from an Aston Villa match in November 2025 based on an intelligence report containing fabricated information generated by Microsoft Copilot, including a match that never occurred and false claims about fan violence in Amsterdam.
What happened
West Midlands Police recommended a total ban on Maccabi Tel Aviv fans attending a UEFA Europa League match against Aston Villa at Villa Park in Birmingham on November 6, 2025. The force's intelligence report classified the event as 'high risk' and cited a history of violence involving the Israeli club's supporters.
The report included details of a '2023 match between West Ham and Maccabi Tel Aviv' where significant disorder allegedly occurred, but this match never took place. The report also conflated and exaggerated accounts of incidents in Amsterdam, claiming 500-600 fans had thrown innocent people into rivers and that 5,000 police officers had been deployed, figures later debunked by Dutch authorities and independent inspectors.
Investigation revealed the 'intelligence' was fabricated by Microsoft Copilot and had been used without adequate verification. Officers, including the police force chief, initially misled Parliament about the use of AI.
Reported impact
Israeli fans were denied the right to support their team based on fabricated information, and the local Jewish community felt alienated when the police force falsely claimed they had supported the ban. An assistant constable later apologized for falsely claiming to have obtained the local Jewish community's approval before enacting the ban.
Chief Constable Craig Guildford resigned on January 18, 2026, stating that 'political and media frenzy' had become detrimental to the force, though he offered no apology for the erroneous policing decision. The suspension of Microsoft Copilot use across the force followed the incident.
Index analysis
Independent reviews cited confirmation bias: police selected intelligence that supported a predetermined decision to recommend a ban rather than weighing all available evidence objectively.
This case illustrates 'shadow AI' (unauthorized use of AI by staff) and 'automation bias', where authorities trust machine-generated data more than human due diligence, resulting in human rights violations.
Governance controls
The incident highlighted the absence of adequate verification procedures and oversight over how AI technologies are integrated into official police processes.
MPs criticized the force, and the UK Home Secretary publicly stated loss of confidence in the police chief.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.