Face palm: Most people already know that there are some tasks that you really shouldn’t rely on AI for, especially those that involve sensitive topics. It should therefore come as no surprise that a British police officer is under investigation for allegedly using Copilot to assist in a criminal investigation.
A senior officer at Derbyshire Constabulary is being investigated by the Independent Office for Police Conduct (IOPC) for alleged gross misconduct. According to the BBC, the investigation began in August after police made a voluntary referral on June 23 following an internal review of the detective’s use of AI.
The allegations go beyond asking Microsoft’s assistant to check some sentences for spelling and grammatical errors. The officer allegedly used AI to formulate policy decisions, assess and manage risks, and justify decisions during criminal investigations. They are also accused of entering sensitive operational and personal data about cases into the tool.
Derbyshire had already approved Copilot for use by officers in May 2025. It should help with tasks such as writing reports and summarizing information. However, this permission was subject to conditions that prohibited the entry of operational, sensitive or personal data.
Emily Barry, the IOPC’s head of engagement, said: “These are serious allegations and it is right that they are investigated thoroughly and independently.”
Investigators will examine the detective’s co-pilot assignment between May 2025 and June 2026 to determine whether professional standards were violated. The officer received a complaint of gross misconduct, but no finding was made.
AI has already found its way into police documents elsewhere. It was previously reported that US police departments were adopting Axon’s Draft One, which generates draft reports from body camera audio. Officials were required to review and approve the results, although concerns remained about inaccurate information and an increasing reliance on automated writing.
Earlier this year there was a bizarre example of AI misreading police paperwork. An AI-generated report in Utah claimed an officer turned into a frog after the software picked up audio of Disney’s “Princess and the Frog” in the background. A human caught the error before the report was submitted.
The element of sensitive data is also reminiscent of Samsung’s experience in 2023, when employees submitted sensitive codes and meeting information to ChatGPT. As reported at the time, attempts to speed up routine work ultimately resulted in company information being leaked to an outside service.
