TL;DR
The Independent Office for Police Conduct is examining allegations of gross misconduct against a Derbyshire detective chief inspector over their use of AI, reportedly in keeping investigation decision logs. Separately, a Gloucestershire police worker has been sacked for repeatedly getting round a force block on ChatGPT. Both cases show forces struggling to police their own use of tools staff have already picked up.
The most senior case so far
The force referred the officer to the watchdog voluntarily on 23 June. That followed an internal review which concluded their use of the technology sat outside the guidance and conditions Derbyshire had issued. The IOPC judged the matter serious enough to investigate and has served the officer with notice. The force says the inquiry is independently led and it cannot say more.
One person familiar with events told the FT the officer holds the rank of detective chief inspector, and that the AI was used to keep decision logs. Those logs are how senior investigators record their reasoning at each step, and defence lawyers can request them for scrutiny at trial. The Crown Prosecution Service has signalled that defence teams in any affected prosecution have been notified.
No more senior officer anywhere in UK policing has been publicly identified as facing an AI-related inquiry. Nor is this Derbyshire’s first: in June the same force opened a criminal investigation into a junior colleague, on a perverting-the-course-of-justice allegation tied to chatbot use while preparing court material.
A block that did not hold
Gloucestershire’s case is the first known sacking of UK police staff over AI. The allegations say the worker used the technology for police business from October 2024 and carried on after the force cut off access to ChatGPT on 3 December that year. The workaround came to light in March 2025, prompting immediate suspension; dismissal followed within the past few weeks. One count alleged confidential material had been compromised, but investigators found everything typed into the chatbot was already in the public domain.
That detail matters. A technical block was in place, went round for more than a year before anyone noticed, and the actual harm proved narrower than charged — a pattern any UK organisation relying on a policy-based AI ban should recognise.
Looking forward
Restrictions on police AI exist to keep fabricated content out of evidence and protect the confidentiality of investigation material. Both cases suggest detection, not rule-writing, is the harder problem. With England and Wales forces already instructed to stop using AI when drafting court statements, the IOPC finding will shape how seriously other constabularies treat their own monitoring gaps.