Deny, deny, admit: UK police used Copilot AI “hallucination” when banning football fans (arstechnica.com)

The chief constable of West Midlands Police has acknowledged that their controversial decision to ban Maccabi Tel Aviv football fans from a match in Birmingham was based on inaccurate information generated by Microsoft Copilot. This admission comes after weeks of denial and highlights significant issues surrounding the use of AI tools in high-stakes decision-making environments. The backdrop of this decision included a recent terror attack in Manchester, which heightened fears of potential violence at the match, prompting the police to recommend the ban based on claims of prior violence by Maccabi fans in Amsterdam. This incident raises important concerns for the AI/ML community regarding the reliability of AI-generated content, often characterized by “hallucinations” where the AI produces fabricated or misleading information. The errors reported, such as the exaggerated number of Maccabi fans involved in violent incidents and the overestimation of police deployment, point to the critical need for human oversight and verification when integrating AI into public safety decisions. The implications of this case extend beyond policing, emphasizing the necessity for transparency and accountability in AI applications, particularly when they impact communities and public perception.