The British Transport Police (BTP) has begun trialing live facial recognition (LFR). Since February 2026, trials lasting several hours have been conducted at major London railway stations including London Bridge, Waterloo, Euston, King’s Cross, St Pancras, Liverpool Street and Victoria. In August, the technology was extended to London Underground stations for the first time.
LFR is now being tested in London’s everyday mass-transit system. The first documented operation took place at Victoria Underground station on 11 August 2026, with further trials at selected stations due to follow by the end of November. The expansion forms part of a broader push to extend facial recognition in Britain, which recently included opening up the driving-license database for police use.
High Error Rates
So far, such programs have proved highly prone to error. The trials in London have likewise yielded no results. The civil-liberties group Big Brother Watch went so far as to describe them as having a “100 per cent fail rate”.
The figures from the first Underground deployment support that criticism. On 11 August, the system operated at Victoria Underground for six and a half hours. The watchlist against which the images were checked contained 673 people, including individuals subject to arrest warrants and others whose compliance with court orders or probation conditions was being monitored.
According to BTP estimates, 33,545 faces were captured during the Victoria operation. The system produced no alerts and no arrests. LFR operations at major London railway stations since 11 February had scanned 538,000 faces and generated just one potential match, which proved to be false.
Including Victoria Underground, around 571,000 faces had therefore been processed by the BTP system by 11 August without a single wanted person being identified.
Any potential match is reviewed by an officer. The algorithm does not itself result in anyone being stopped. An officer must compare the image with the person on the ground and decide whether further checks are warranted.
The scale of the operation also raises privacy concerns. BTP says images and biometric data relating to people who do not trigger an alert are deleted immediately and automatically. Material linked to a potential match is deleted after use or within 24 hours at the latest.
The system also records conventional CCTV footage, which may be retained for up to 31 days. Transport for London, BTP says, has no access to either the biometric data or the watchlist.
Changing the Nature of Policing
The criticism is not confined to the software’s susceptibility to error. More fundamentally, it concerns a change in the way police searches are conducted. Traditionally, police use established investigative methods to look for a specific person. With LFR, by contrast, large numbers of people who are not suspected of any wrongdoing are subjected to biometric checks in the hope of identifying a wanted person. Policing thus shifts from targeted searches toward a system driven by the mass processing of personal data.
Britain has no law expressly granting police the power to use LFR. BTP bases its authority on a patchwork of common law, the Police and Criminal Evidence Act, data-protection law, the Human Rights Act and various guidelines and court rulings. BTP itself acknowledges that Article 8 of the European Convention on Human Rights, which protects the right to private life, is engaged.
A Shaky Legal Basis
The UK’s data-protection regulator, the Information Commissioner’s Office, does not regard facial recognition as inherently unlawful, but requires a clear legal basis and adequate safeguards. The regulator has also called for a more specific statutory framework governing police use of the technology.
Critics argue that a system capable of biometrically checking tens of thousands of passers-by within hours should not be introduced without legislation specifically enacted to govern its use.
This is not the first step of its kind. The Metropolitan Police has previously deployed live facial recognition on London streets and at certain crime hotspots. There, too, the police were operating on legally thin ice. Extending it to the Underground now brings the technology into a part of everyday life used by millions of people.
China Shows Where This Can Lead
Civil-liberties critics accordingly warn that facial recognition could evolve from an exceptional law-enforcement tool into a routine feature of public infrastructure. Big Brother Watch argues that a person’s face could effectively become a permanent form of identification that can be checked anywhere and at any time. With comprehensive LFR, everything from tracking people’s movements to scrutinizing their political views, for example by recording attendance at rallies and demonstrations, becomes possible.
China offers a stark example of where such systems can lead. With cameras monitoring virtually the entire public realm, the distance from surveillance on every street corner to a “social score” is short. It begins when the state starts automatically identifying people as they go about their daily lives.
The technology now being trialed in London provides the essential technical foundation for linking behavior, identity and government-held data on a comprehensive scale.