A police officer in Derbyshire, England, is under criminal investigation for allegedly using artificial intelligence to create false evidence in multiple cases, Sky News reported on 13 June 2026. Prosecutors are helping with the inquiry. No arrests have been made.

The officer has been taken off active duty while the investigation continues. If found guilty, the charge would be perverting the course of justice — a serious crime that reflects how seriously the courts treat tampering with evidence in criminal trials.

Prosecutors are involved early in the investigation, which is notable. They typically advise police before charges are brought, helping to shape the inquiry in ways that will hold up in court. Their presence here suggests investigators are already checking whether specific cases or convictions were affected by the alleged fabrication.

We do not yet know which specific AI tools were used. That matters because different tools create different problems. Text-based AI can generate realistic-sounding witness statements. Image AI can create fake photographs or footage that looks like CCTV. Audio AI can mimic someone's voice. Each type requires a different way to detect and investigate.

Finding AI-generated fake evidence in old case files is a challenge. Most detection methods are designed to work on new content going forward, not to catch fakes that were already hidden in case records. The tools that try to identify signs of AI creation are not yet perfect — they can produce false alarms or miss real fakes, and they become less reliable as AI systems improve. Any review of this officer's cases will require careful, manual work rather than a simple automated scan.

The UK police force has faced sustained public and judicial scrutiny around how evidence is handled. The Post Office Horizon scandal revealed what happens when institutions allow systemic problems to go unchecked. This case is different — it involves one officer acting deliberately — but the damage to defendants is the same: their convictions are based on fabricated material.

Regulatory bodies have warned about this kind of problem in recent years. The College of Policing has been drafting guidance on how law enforcement can use AI responsibly, and police leadership has worked to define what is and is not acceptable. Those guidelines were built mainly around predictive algorithms and facial recognition — not the deliberate misuse of generative AI by individual officers. This case reveals a gap in those protections.

Because the investigation involves multiple cases, the consequences could extend far beyond any punishment for the officer. People who were convicted in cases this officer worked on will have legitimate grounds to appeal their convictions through the Criminal Cases Review Commission. That commission already has a large backlog; a surge of appeals tied to one officer's conduct could strain its limited resources.

Derbyshire Police has not yet said how the suspected fabrication was discovered, when it started, or how many cases are affected. Those details will be important — both for the prosecution and for understanding how much damage control is needed.

Criminal justice depends on the assumption that evidence presented in court is real. AI can now create convincing fake material very cheaply and without much technical skill. This changes the risks — not just from outside criminals but, as this case shows, from people inside the system.