Derbyshire police officer investigated over alleged AI-generated evidence

Investigation follows formal launch of new PoliceAI centre

An officer in Debyshire police has been suspended from frontline duty and is subject to a criminal investigation for allegedly using AI to create evidence.

Derbyshire police have launched a criminal investigation into an officer for allegedly using AI to create evidence and perverting the course of justice.

It is thought to be the first known case of its kind in the UK.

Derbyshire police told the Financial Times: “A criminal investigation has been launched into an allegation of perverting the course of justice after the alleged use of AI systems by an officer to create evidential material in a number of cases.

“The force is working closely with the Crown Prosecution Service in relation to any potentially impacted cases.”

The force added the investigation was “in its early stages” and no further details were available.

It said: “The officer involved has been removed from frontline duties, pending the outcome of the investigation. No arrests have been made.”

The officer’s role or the exact nature of the suspected misconduct has not been disclosed.

A Crown Prosecution Service spokesperson said in a statement to The Guardian: “We are working with Derbyshire police as it conducts enquiries into the alleged use of artificial intelligence by an officer.

“We are engaging with defence teams and the courts in appropriate cases. As police enquiries continue, it would not be appropriate to comment further.”

Risks of hallucinations

The investigation comes after the formal launch last week of PoliceAI, a new national centre for AI in policing. The centre evolved from the National Police Chiefs’ Council’s (NPCC) AI portfolio and the goal is to free officers from bureaucratic burdens in order that they can dedicate more time to frontline policing and bringing offenders to justice faster.

However, Alex Murray, Head of PoliceAI, said last week he had intervened in some cases to temporarily halt the use of generative AI for this sort of work.

'We need to slow it down a bit,' Murray said, adding that technologies used within the criminal justice system must achieve a standard of accuracy that is 'beyond reasonable doubt.'

LLMs do still make a lot of mistakes and authorities fear (wisely) that even minor inaccuracies generated by AI could contaminate evidence, mislead courts or undermine confidence in criminal proceedings.

The most famous example of this probably occurred earlier this year when the Chief of West Midlands police had to apologise to MPs for giving them incorrect evidence about a decision to ban Maccabi Tel Aviv fans from a Europe League match, saying it was the result of a mistake by Microsoft Copilot which apparently made up a match between Maccabi Tel Aviv and West Ham and that mistake informed the police decision to ban fans from the match.

Last week at London Tech Week, Deputy Prime Minister David Lammy announced the launch of AI legal assistants for use in Crown Court cases, alongside an AI tool for judges to identify cases ready for trial. In a fireside chat with Financial Times Tech New Editor Murad Ahmed, Lammy acknowledged that generative AI tools had made mistakes but suggested that the impact of these mistakes could and should be mitigated. He said:

“I think it's about staying at the appropriate level and about looking at that data to get it right and trying to ensure that the consequences where there can be hallucination error are not overly significant. The consequences of it being wrong are too much for any society to comprehend.”

Lammy also emphasised the importance of transparency about the use of AI in the legal system.