Artificial intelligence tools are being introduced across policing and the criminal justice system in England and Wales more rapidly than the regulations meant to oversee them, according to new research from Northumbria University. The study mapped 70 AI tools already in use, under trial or in development. These include systems for contact center triage, drafting witness statements, digital forensics and facial recognition. The four-year project involved partnerships with several other universities and was funded by Responsible AI UK. Researchers built an interactive mapping tool and interviewed police officers, officials, legal experts and technology providers between May 2025 and January 2026. Of the tools identified, 27 are operational and 34 are in pilot stages. More than half originate from commercial vendors, with most activity focused on community policing, intelligence and investigations. The study found that AI delivers benefits in areas such as transcription, crime analysis and officer welfare when carefully designed and evaluated. However, adoption is advancing faster than supporting governance and evidence frameworks. The principle of keeping a human in the loop often provides only nominal oversight. The report also notes that highly accurate systems may reduce human checks, allowing rare errors to go unnoticed, especially when tools are connected in sequence. The authors issued 26 recommendations for improved evaluation, transparency and training. The findings coincide with new national efforts to strengthen AI coordination in policing.
