British Transport Police spent £320,786 on an 18-deployment live facial recognition trial across London's busiest railway stations between February and July 2025. The system scanned more than half a million faces. It produced a single watchlist alert — a false positive. No arrests resulted from the technology itself. The numbers are stark: roughly £320,000 per zero correct identifications. Nearly 100 hours of police officer time. Eighteen separate deployments across six months. The cost-per-outcome ratio is effectively infinite because the denominator is zero. BTP notes that officers made "a number of associated arrests" during deployments — for assault, theft, weapons possession — but concedes these "did not result directly from an LFR alert." In other words, police standing visibly in stations caught criminals the old-fashioned way, while a £320,000 camera system whirred uselessly beside them. Despite this record, BTP has extended the trial four more months and expanded it to London Underground stations. Since the extension, three confirmed alerts have occurred — but all three were people already complying with court orders, not fugitives caught in the act. Transport for London endorsed the expansion by framing it around "violence against women and girls," a politically potent justification that sidesteps the question of whether the technology actually delivers on that promise. The broader rollout is accelerating regardless. More than half of police forces in England and Wales have now deployed LFR on streets. London Mayor Sadiq Khan confirmed fixed cameras will be installed on the newly pedestrianised Oxford Street. The Metropolitan Police and the mayor's office announced permanent West End installations. Each expansion cites the same justification — public safety — while the evidence base from the one trial that produced auditable numbers shows precisely nothing. Fraser Sampson, former UK biometrics and surveillance camera commissioner, noted the gap between retail and policing applications: success in a shop means deterrence, success for police means catching people. By that standard, the trial failed on its own terms. Sampson now works for Facewatch, a facial recognition company operating in shops, which adds a useful disclosure but does not diminish his structural point about proportionality. Parliament's Joint Committee on Human Rights flagged LFR rollout as a "particularly clear example of risk" to human rights and called for legislation. Labour MP Bell Ribeiro-Addy called for suspension of the rollout pending a proper legal framework. No such framework exists. The Information Commissioner's Office has flagged oversight concerns but has not blocked deployments. The pattern here is familiar: a surveillance technology is trialled, produces negligible results, is expanded anyway, and accumulates institutional momentum before any legal framework catches up. BTP's own explanation — that the watchlist was deliberately kept small while "governance and safeguarding arrangements" were established — is either a reasonable explanation for low alert volume or a retrospective justification for poor results. The distinction matters, because the infrastructure being built will outlast any single trial's parameters.
