FOI docs: UK police's six-month live facial recognition trial in London railway stations scanned 500K+ faces, leading to no arrests and one false-positive alert

Police live facial recognition trials have yielded no arrests and one false positive, costing over £300,000.

A six-month trial of live facial recognition (LFR) technology by British Transport Police (BTP) in London railway stations concluded with no arrests and a single false positive identification. The trial, which ran from February to July and cost over £320,000, scanned more than 500,000 faces across busy transport hubs. The technology was intended to help identify offenders and individuals breaching court orders. Despite the limited success in terms of direct alerts, BTP has stated that associated arrests for offenses like assault and theft were made during the deployment periods, though these were not directly linked to LFR matches. BTP is extending and expanding the trial to include London Underground stations, aiming to target violence against women and girls.

The underwhelming results of the BTP's LFR trial, with no arrests directly attributed to the technology and a single false positive out of over 500,000 scans, raise significant questions about its efficacy and cost-effectiveness. Despite the official statement about associated arrests, the lack of direct LFR-generated successes highlights potential limitations in the technology's current application for public transport surveillance. This outcome could influence public and governmental perceptions, potentially impacting future funding and regulatory decisions regarding widespread LFR deployment by law enforcement.

The continued expansion of LFR trials by BTP and the Metropolitan Police, particularly with stated aims of tackling violence against women and girls, signals a persistent governmental push for this surveillance technology despite documented performance concerns. The trial's cost and low yield of direct alerts contrast with the broad deployment of LFR across over half of UK police forces. This divergence suggests a market where technological adoption is being driven by broader strategic goals rather than solely by demonstrated operational success, setting a precedent for future surveillance infrastructure investments and ethical debates.