UK Police Told to Stop Using AI for Court Statements Over Concerns of False Content
2026-06-08 09:44
Favorite

en.Wedoany.com Reported - Multiple UK police forces have been instructed to stop using artificial intelligence systems to draft court statements and carry out other criminal justice-related work, due to concerns that inaccurate AI-generated content could disrupt judicial processes.

Concerns over AI-generated false content disrupting justice lead UK police to halt use of AI for court statements

Alex Murray, head of the UK Police AI Centre (Police.AI), said he had intervened with some forces that had begun using commercially available AI tools without completing a full compliance assessment. Murray told the affected forces: "Stop immediately. We need to slow down the pace." He added that any technology used in the criminal justice system must meet the rigorous standard of accuracy beyond a reasonable doubt.

The suspension highlights growing controversy over the use of generative AI in policing. Murray, who oversees AI use by police forces in England and Wales, believes the technology could ultimately speed up case investigations and evidence processing, but robust safeguards must be established before widespread deployment.

Police.AI was officially established this year as part of UK government reforms to improve overall policing efficiency. Home Secretary Shabana Mahmood said the project has received £115 million (approximately 1.046 billion yuan) in funding over three years, and the time saved by its implementation is equivalent to adding 3,000 officers to the existing workforce of around 145,000 in England and Wales.

According to Murray, some forces began using commercial AI systems to help officers convert interview transcripts into court statements, prompting intervention by the Police AI Centre. Murray did not name the specific forces involved, but said police had also been reminded not to apply AI to other policing tasks until full compliance reviews were completed. "We made it clear to some forces: you can't do this yet because the full set of checks and balances hasn't been completed," he said.

Murray specifically noted that the preparation of disclosure lists requires extra caution. These lists are records of evidence that must be submitted to the defense before trial. Forces using AI for this task must explain how the technology was trained and how it is applied in practice. Murray acknowledged that police have had previous experiences with AI. Last year, West Midlands Police used content generated by Microsoft's Copilot assistant while compiling a ban order application. The document, which sought to ban a football club's fans from attending matches, fabricated a past match involving Israel's Maccabi Tel Aviv, which the fans were originally scheduled to attend against Aston Villa. This incident further raised widespread concerns about AI hallucinations—where AI systems generate information that appears real but is actually false. "All forces now have policies in place for the proper use of Copilot, requiring officers to verify all content generated by the tool," Murray said.

Despite the risks, Murray believes that thoroughly tested AI systems could bring significant benefits to forces struggling with surging workloads and increasingly complex investigations. Video surveillance analysis is a promising application. Currently, officers often spend hours manually reviewing footage to identify suspects. Murray said: "With the right instructions, AI can search as required, for example, 'find the man in the red hat.' Within the critical golden hours or days when evidence still holds key value, we can locate the target, make an arrest, and seize blood-stained shoes at the scene." He also noted that AI can handle vast amounts of electronic evidence, such as in child sexual abuse investigations. The Police AI Centre is currently developing a tool to identify and categorize images on seized devices, reducing the need for officers to directly view harmful content. "Officers only need to sample-check rather than view every image. Such content can cause immense psychological harm. I believe that with proper regulatory rules, usage guidelines, and personnel training, the benefits of automation far outweigh the drawbacks," Murray said.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com