England and Wales Police Seek AI Safeguards for Digital Evidence

Police forces across England and Wales are exploring artificial intelligence to process growing volumes of digital evidence from mobile phones, cloud services, social media, connected devices, recorded calls and video. The potential procurement driver is operational pressure: investigators, particularly in smaller teams with limited resources and budgets, can face hours of footage and large quantities of messages that are difficult to review manually. AI-assisted search and analysis could help reduce that workload, but adoption is being slowed by uncertainty about liability, data protection, evidentiary integrity, admissibility in court and the lack of consistent national guidance.

The source frames safeguards as essential to realizing the efficiency gains without weakening investigations. John Lucey, VP EMEA at digital forensics company Cellebrite, cautioned that policing decisions carry consequences “that most commercial AI tools were never built to withstand.” He warned that rushing deployment could create evidence weaknesses and cause cases to collapse. For government buyers, this points toward investigation-specific systems whose outputs can be traced and scrutinized—not simply general-purpose AI features. Requirements may need to address transparent processing, audit logs, secure handling of sensitive evidence, continuous testing and clear responsibility for reviewing and acting on AI-generated findings.

Lucey described an example involving a drone used to deliver drugs into a prison. Investigators recovered an SD card from the destroyed device containing hours of footage. AI identified a test flight in a living room and extracted details from a hospital appointment notice visible beneath a television, including the appointment date, hospital and doctor’s name. After verifying the information, investigators had the suspect in their office within an hour, according to Lucey. He also cited AI searches of recorded prison phone conversations for references to drugs or firearms, reducing work that could otherwise take dozens of hours to minutes. These examples illustrate potential mission value, but they also make validation, provenance and human review central procurement concerns.

The signal is market intelligence rather than a live acquisition notice: it identifies no specific solicitation, contract value or award. Buyers can use the concerns described to shape market research, evaluation criteria and pilot conditions, including how suppliers demonstrate evidentiary integrity, data-protection compliance, testing practices and accountable human oversight. Lucey summarized the intended role of the technology: “AI is an assistant to accelerate casework and connect siloed data, not an investigator – it helps us get to the truth faster, but it doesn’t decide the outcome of a case.” The source also promotes registration for the Think AI for Government conference in October 2026, but gives no specific event date.

• Operational need: Police are assessing AI to address growing digital-evidence backlogs across footage, messages and other data.
• Uneven capacity: Smaller teams with modest resources may have particular difficulty reviewing hours of material manually.
• Evidence risk: The source warns that rushed deployment could weaken evidence and potentially contribute to cases collapsing.
• Procurement criteria: Consider traceable outputs, auditable logs, secure data handling, ongoing testing and documented human review.
• Data protection and admissibility: Suppliers should explain how systems protect sensitive information and support evidence that can withstand scrutiny in court.
• Illustrative use case: AI analysis of drone SD-card footage surfaced a hospital appointment notice; investigators verified the information and located the suspect within an hour, according to Lucey.
• Additional application: Searching recorded prison calls for references to drugs or firearms could reduce manual review from dozens of hours to minutes.
• Human accountability: Cellebrite’s John Lucey characterized AI as assisting casework and connecting siloed data—not deciding case outcomes.
• Market status: No named procurement, solicitation, contract value or award is provided; the signal is an indicator for requirements development and market research.
• Event mention: The source promotes the Think AI for Government conference in October 2026 without specifying a date.