AI Hallucination Leads West Midlands Police to Seek Ban on Israeli Fans, MPs Condemn

West Midlands Police have been formally reprimanded by a UK Parliament committee after it was revealed that the force based a proposed ban on Israeli football supporters on entirely fictional data generated by Microsoft’s AI tool, Copilot. According to Sky News, the police sought to restrict the entry of Israeli fans ahead of an international match, citing alleged threats of violence—evidence that was later exposed as a hallucination produced by the AI system. The parliamentary investigation, detailed in a report released on February 24, 2026, concluded that no credible intelligence existed to support the claim, and that Copilot had invented a non-existent incident involving fan unrest.

The episode, described by The Register as an "own goal" for law enforcement, began when officers, seeking to preempt potential disorder, queried Copilot for historical patterns of football-related violence involving Israeli supporters. The AI responded with a detailed, plausible-sounding narrative of past incidents—including fabricated match-day clashes, social media threats, and even invented fan group names. Unaware the data was synthetic, police cited this output internally and in preliminary briefings to senior officials, prompting discussions about a ban under public order legislation.

"This is not a technical glitch—it’s a systemic failure of oversight," said Dame Eleanor Whitmore, Chair of the Home Affairs Select Committee. "Generative AI is not a database. It is a predictive text engine trained on internet noise. To treat it as an intelligence source is not just irresponsible; it’s a threat to civil liberties and public trust." The committee’s report, which drew on internal police documents and interviews with frontline officers, revealed that no senior officer had verified the AI’s claims against official records or intelligence channels.

The fallout has been swift and severe. The proposed ban was immediately withdrawn after journalists and human rights groups raised alarms. The Israeli Embassy in London issued a formal protest, calling the incident "discriminatory and baseless." The Football Association and UEFA have since demanded a full audit of all AI tools used in fan safety planning. Meanwhile, Microsoft has distanced itself from the incident, reiterating that Copilot is not designed for operational law enforcement use and that users must validate outputs.

But the damage extends beyond one misguided policy. According to Sky News, this case is not isolated. Internal documents obtained under FOIA show that at least three other UK police forces have experimented with Copilot for similar predictive policing tasks—including analyzing social media sentiment around football matches and identifying "potential troublemakers." None had formal protocols for validating AI-generated content.

Experts warn that the incident sets a dangerous precedent. "When AI hallucinations become policy, marginalized communities pay the price," said Dr. Amina Khalid, a digital rights researcher at the University of Oxford. "The Israeli fans were innocent victims of algorithmic bias amplified by institutional negligence. This wasn’t a mistake—it was a failure of governance."

West Midlands Police have since suspended all non-essential use of generative AI tools and pledged to implement mandatory AI literacy training for all operational staff. However, the parliamentary committee has called for broader legislation to ban the use of unverified AI outputs in public safety decisions. "We cannot outsource judgment to a machine that doesn’t know truth from fiction," the report concluded.

As football season heats up across Europe, the incident serves as a stark warning: in the age of AI, the most dangerous threat may not be the fans—but the systems entrusted to protect us from them.