British Transport Police have been trialling live facial recognition (LFR) cameras at several of London’s busiest train stations as part of a six‑month pilot that began in February.
The pilot has operated at six stations—Waterloo, Euston, Victoria, King’s Cross, St Pancras and London Bridge—scanning an estimated 330,000 faces. Individual counts are roughly 92,800 at Waterloo, 71,900 at Euston, 51,900 at Victoria, 45,300 at King’s Cross, 17,500 at St Pancras, 17,200 at Liverpool Street and 16,200 at London Bridge.
Only a single alert was generated, at King’s Cross, and it was later identified as a false alarm; no confirmed matches have been recorded to date.
Police officials say the system is calibrated with a deliberately high threshold to minimise misidentification, and every alert is subject to officer review before any further action is taken.
During the deployments officers made 13 arrests for offences such as breach of criminal‑behaviour orders, theft and public‑order violations, and four individuals were identified as wanted by other forces. These arrests were not directly linked to facial‑match alerts, but police note a reduction in crime during periods when the cameras were active.
A civil‑rights group warned that the pilot raises serious concerns about privacy and the risk of innocent commuters being subjected to intrusive identity checks without their knowledge.
Live facial recognition technology is also being used by private operators in retail environments, where documented cases have shown shoppers being mistakenly accused of theft.
In one incident, a shopper in Woolwich was publicly labeled a thief at a sports‑wear retailer, and another was escorted out of a supermarket in Elephant and Castle after a false identification.
The Metropolitan Police have expanded their own permanent LFR programme to areas such as the West End and Soho after reporting 173 arrests during a six‑month trial in Croydon, including the capture of a woman wanted for more than two decades.
Residents in locations where the technology has been trialled have expressed mixed reactions, with some citing perceived reductions in crime while others question its effectiveness against individuals wearing face coverings.
Live facial recognition works by capturing images of people as they pass a camera, streaming the footage to a database, and comparing it against a watchlist to enable rapid identification when a match is found.
The ongoing debate balances potential security benefits against privacy risks as authorities consider broader deployment of the technology across the city.
