The British Transport Police recently concluded a trial of live facial recognition technology across 18 London railway stations. Despite scanning half a million people and spending over £300,000, the initiative failed to produce a single direct arrest.
Zero arrests for £321,000 in spending
The British Transport Police (BTP) deployed a system known as Live Facial Recognition (LFR) in a three-month experiment running from February to July. According to a report by The Guardian, the operation cost approximately £321,000 and consumed nearly 100 hours of officer time. While the BTP scanned more than 500,000 faces, the technology did not lead to any arrests directly linked to the LFR alerts.
The system did trigger one match against a police watchlist, but that alert was ultimately identified as a false positive.. While the BTP did record several arrests for crimes including theft and assault during the trial period, the agency clarified that these were the result of standard policing duties rather than the LFR technology. This distinction highlights a significant gap between the operational cost of the surveillance and its actual utility in apprehending suspects.
A four-month extension into the London Underground
Despite the lack of tangible results, the British Transport Police are proceeding with a four-month extension of the LFR trials. This next phase will move the technology into the London Underground network, a move backed by Transport for London (TfL). TfL has indicated that the goal is to identify individuals on police watchlists at high-impact statinos, specifically to protect women and girls from violent offenders and sexual harassment.
The expansion comes amid mixed internal reviews. Máire Thierry, the commissioner of the Police Service of Northern Ireland, suggested that LFR trials have previously identified individuals sought by courts and led to "associated arrests" for other crimes. However, the BTP's specific London trial has yet to mirror that success, leaving the upcoming Underground tests as a critical pivot point for the program's viability.
Fraser Sampson's warning on the 'watchlist' limitation
The efficacy of the BTP's approach has been questioned by industry experts, including Fraser Sampson, a former UK biometrics commissioner.. Sampson, who now works for the retail-focused provider Facewatch, argued that facial recognition functions differently in transit hubs than in shops. While a retail store can use the tech to deny entry to a known troublemaker, the same logic does not necessarily help police catch a suspect once they have vanished into a crowded community.
Sampson emphasized that LFR is entirely dependent on the quality and breadth of the watchlists provided by the police. as reported in the source, if a suspect is not already on a specific list, the system will neither flag them nor retain their image. This creates a circular dependency: the system cannot find "new" criminals, only those the police have already identified and cataloged, which may explain the zero-arrest outcome of the initial 500,000 scans.
The 250 million face scale of US federal surveillance
The scale of the BTP trial is dwarfed by surveillance efforts in the United States, where federal agencies like the Department of Homeland Security and the Internal Revenue Service use systems capable of scanning 250 million faces in near real-time. This comparison underscores a global trend toward mass biometric scanning, though it also highlights the differing regulatory environments between the UK and the US.
Civil-rights advocates in both nations remain concerned about the lack of transparency regarding who is placed on watchlists and how that data is managed. In the UK, the BTP spent nearly £200,000 on equipment and training in the first half of the year alone—a cost that would double if the system were rolled out across all London stations.. The central question remains whether the BTP can prove LFR is a reliable tool or if it is simply an expensive addition to an already saturated CCTV landscape.
Comments 0