Police officers risk accidental AI misconduct, watchdog warns

Police officers and staff could face misconduct investigations or dismissal for using artificial intelligence (AI) without fully understanding the rules governing its use, the Independent Office for Police Conduct (IOPC) has warned.

The watchdog has written to the National Police Chiefs’ Council (NPCC), urging forces to review their AI policies and ensure officers receive clear guidance and practical training.

The warning follows allegations that officers and staff have used AI tools inappropriately or outside the limits of their force’s policies.

The IOPC confirmed on 13 October that it is conducting two independent investigations into allegations of gross misconduct involving AI. Other cases have been investigated or concluded locally by police forces.

IOPC Director General Rachel Watson said officers attempting to improve efficiency could inadvertently breach professional standards.

“There is a risk that police officers and staff, believing they are doing the right thing by using AI, perhaps to speed up a process, could end up being investigated for misconduct because they don’t know or understand what’s allowed,” she said.

Police AI misuse investigations

The IOPC’s warning follows several investigations into the use of AI by police officers and staff.

As reported by the BBC in August, the watchdog opened a gross misconduct investigation into a senior Derbyshire Police detective who allegedly used AI to maintain investigative decision logs, potentially breaching the force’s guidance. 

ITV reported that a separate Derbyshire Police officer was placed under criminal investigation in June over allegations that AI was used to create evidential material for court proceedings.

The Financial Times also reported that Gloucestershire Police dismissed a staff member who repeatedly bypassed restrictions on ChatGPT to use the technology for police work.

An HMICFRS inspection identified officers at Thames Valley Police using external AI applications to prepare reports and witness statements, breaching data security requirements.

Police forces urged to strengthen AI policies

Watson called for all police forces to establish robust AI policies that clearly explain permitted uses of the technology.

She said guidance must be regularly reviewed to reflect technological developments and communicated effectively to officers and staff.

“And AI policies need to be effective, not just a tick box, designed around operational reality with clear day-to-day procedures in mind,” Watson said. “Leadership commitment on AI needs to be visible and training built around how to apply AI in practice, not simply awareness of AI systems and conditions.”

The IOPC has raised its concerns directly with the NPCC. Watson has also met the national policing lead for AI.

In an interview with the Financial Times, Watson warned that inaccurate or fabricated AI-generated information entering criminal proceedings could undermine prosecutions and potentially result in mistrials.

Helping police adopt AI ‘effectively and safely’

The IOPC said its two ongoing independent investigations concern allegations of gross misconduct. It has not provided details of the cases in its latest statement.

The watchdog also acknowledged that some allegations involving AI have been handled directly by police forces without being referred to the IOPC.

Watson said the technology presents unfamiliar challenges for professional standards departments, which may be uncertain whether particular allegations require independent investigation.

“These kinds of allegations may be new territory for professional standards departments in some police forces, and it could lead to uncertainty over whether some AI-related allegations warrant a referral to us,” she said. “I would say that if there is any doubt, then it is prudent to refer the matter to us so that our specialist assessment team can review the circumstances and offer advice.”

The IOPC stressed that it supports the responsible adoption of AI in policing and does not want concerns about misconduct to prevent forces from using the technology.

“There is no doubt AI can bring enormous benefits to policing,” Watson said. “Our aim is not to discourage innovation, but to help policing adopt AI effectively and safely, enabling officers and staff to use these technologies with confidence and reducing the risk of avoidable mistakes that could lead to unnecessary involvement in accountability processes.”

She added that the public must also be confident that police will use AI fairly, transparently and with appropriate safeguards.