UK detective faces investigation for using AI in criminal cases

Allegations of misconduct

A senior detective with the Derbyshire Constabulary is currently under investigation for alleged gross misconduct. The Independent Office for Police Conduct (IOPC) began its inquiry in August following a voluntary referral from the police force. This referral came after an internal review discovered the detective's specific methods for using artificial intelligence.

The allegations suggest that the officer went far beyond using the tool for simple administrative tasks like checking spelling or grammar. Instead, the detective reportedly used Microsoft Copilot to formulate significant policy decisions and manage risks during criminal investigations. They are also accused of using the tool to justify specific decisions made during these active cases.

The investigation aims to determine if the officer breached professional standards by relying on automated systems for high-stakes work. While the officer has received a gross misconduct notice, the IOPC has not yet made a final finding. Emily Barry, the IOPC's director of engagement, stated, "These are serious allegations and it's right they are investigated thoroughly and independently."

Violating internal rules

The Derbyshire Constabulary had officially approved the use of Microsoft Copilot for its officers in May 2025. The police force intended for the tool to assist with routine paperwork, such as summarizing large amounts of information or drafting standard reports. These tasks are generally considered low-risk and suitable for automation in a professional setting.

However, the permission granted to the force came with very specific and strict conditions. Officers were explicitly told that they were barred from inputting any operational, sensitive, or personal data into the system. The goal was to ensure that private information regarding citizens or ongoing crimes remained secure within the police infrastructure.

The detective in question is accused of ignoring these boundaries by entering sensitive operational and personal information into the tool. By feeding this data into an external AI assistant, the officer may have exposed confidential details. This breach of protocol is a primary focus of the ongoing inquiry into the detective's conduct.

The risks of AI in policing

The use of AI in law enforcement is already expanding across different jurisdictions and departments. For example, some US police departments have adopted Axon's Draft One, a tool designed to generate report drafts from audio recorded on body cameras. In those instances, officers are required to review and approve every result produced by the software to ensure accuracy.

Despite these safeguards, concerns persist regarding the accuracy of automated writing and the potential for human over-reliance on technology. There have been documented instances where AI has produced bizarre errors in police paperwork. In one case in Utah, an AI-generated report claimed an officer turned into a frog because the software misinterpreted audio from a movie playing in the background.

While a human caught that specific mistake before it was submitted, the incident highlights the unpredictability of these systems. When AI is used to summarize evidence or draft reports, the risk of hallucination—where the AI creates false information—remains a significant hurdle for law enforcement agencies. These errors can have serious consequences if they are not caught by a human reviewer.

Data privacy and security

The incident in the UK mirrors broader concerns about how employees handle confidential information when using popular AI tools. In 2023, a similar situation occurred at Samsung, where employees submitted confidential code and meeting details to ChatGPT. This happened as staff attempted to use the tool to speed up routine work tasks, but it resulted in company information being handed to an outside service.

These examples illustrate a common trend where the desire for efficiency leads to the accidental sharing of proprietary or private data. When information is entered into a public AI model, it may be used to train the system or be accessible to the service provider. For organizations dealing with sensitive criminal data or trade secrets, this creates a major security vulnerability.

The Derbyshire Constabulary investigation serves as a case study on the importance of clear guidelines for AI usage. It shows that even when a tool is officially approved for a workplace, the specific types of data allowed must be strictly controlled. Without these boundaries, the benefits of AI can be overshadowed by the risks of data leakage and compromised privacy.

The path forward

Investigators are currently looking at the detective's use of Copilot between May 2025 and June 2026 to see if professional standards were breached. This period of review will help determine the extent of the data exposure and the impact on the criminal cases involved. The findings will likely influence how other police forces manage their own AI deployments and training programs.

As AI continues to integrate into government and law enforcement workflows, the focus will likely shift toward more robust oversight mechanisms. Agencies may need to implement more sophisticated