Crime-fighting AI will carry risks of bias, says policing lead

But police are in ‘an arms race’ with criminals

The UK’s AI policing lead had admitted that systems will contain bias at launch, but a new dedicated centre will work on improvements over time.

National Crime Agency director Alex Murray said the new £115m National AI Centre would work to reduce unfairness and improve oversight of AI systems that might contain elements of bias.

The senior officer, who leads on AI for policing, said forces across England and Wales were engaged in an "arms race" with criminals who were increasingly exploiting the same technology.

"If you talk about live facial recognition or predictive policing, there will be bias," Murray told The Guardian.

"You need data scientists and engineers to clean the data, to train the model appropriately, and then to test it.

"There is no point releasing something to policing that has bias in it that's not recognised, and everything should be done to minimise it to a level where it can be understood and mitigated."

The admission follows mounting scrutiny of facial recognition systems used by UK police forces.

Retrospective facial recognition, where suspects’ images are compared against the Police National Database after a crime, has already shown evidence of bias.

A December report commissioned by the Home Office and conducted by the National Physical Laboratory found the system was significantly less reliable for certain ethnic groups and genders.

Live facial recognition (LFR), which scans faces in real time to identify wanted individuals, has proved more controversial. While used more sparingly, LFR has also been shown to contain bias.

The Association of Police and Crime Commissioners (APCC) said problems with facial recognition systems had been understood for some time, and criticised the lack of transparency with affected communities.

Under plans backed by Labour, policing in England and Wales will see a dramatic expansion in the use of AI, including dozens of new live facial recognition vans.

National AI Centre works on centralising police AIt

Murray said the new national AI centre would aim not only to reduce bias but also to streamline procurement and testing of technology.

Currently, individual forces make their own decisions about the private-sector AI products they adopt, which critics say is slow and wasteful. The Centre would evaluate tools centrally and ensure safeguards are built in before systems are rolled out widely.

Murray argued that AI's benefits extend well beyond forecasting crime.

He pointed to cases where criminals had attempted to weaponise the technology themselves.

In one instance, a paedophile claimed images of child abuse were deepfakes, forcing investigators to prove their authenticity in court.

AI is also being used to sift vast quantities of digital evidence.

In one recent investigation, four suspects from Luton were arrested over attacks on cashpoints. Officers downloaded data from their mobile phones, much of it in Romanian. AI tools translated and analysed the material, identified relevant evidence and presented it in a format detectives could use, leading to guilty pleas within weeks.

Trevor Rodenhurst, chief constable of Bedfordshire Police, said the technology enabled officers to extract evidence from multiple devices containing huge volumes of data that would otherwise have been impossible to process quickly.

Murray added: "What took days, weeks, sometimes months can potentially take hours."

He stressed, however, that human officers would retain responsibility for final decisions.