Static Facial Recognition is Coming to London's busiest districts
The MET just announced it’s rolling out static live facial recognition (LFR) cameras across London’s West End and Soho by December 2026. Not in the experimental phase—not yet. This is a full deployment built on six months of operational testing in Croydon. They’re attaching cameras to lampposts and other fixed infrastructure, feeding the feeds to remote monitoring stations where ground officers can physically stop people flagged by the system.
Here’s what that looks like in practice: when officers set up a camera, they upload a bespoke watchlist—maybe five names, maybe ten—to match against. The list gets deleted within 24 hours after the deployment wraps up, according to Metropolitan Police documentation. But here’s the thing: even with that delete-after-use policy, civil liberties groups are screaming foul. Big Brother Watch calls it "permanent biometric surveillance of the public square" and says this flies in the face of policing by consent.
I’m not saying you should panic. I am saying you should pay attention—because this isn’t just about one stop-and-notice moment. This is about who gets watched, why, and how often—especially when the technology has a documented history of racial bias leading to false matches.
How Croydon’s pilot laid the groundwork for city-wide LFR
The last six months in Croydon serve as the proof-of-concept for this expansion. Between October 2025 and March 2026, the Met ran 24 LFR deployments across central Croydon. They didn’t just point cameras at random pedestrians and see what happened. Each operation used a custom watchlist tailored to the specific threat profile for that deployment—sometimes drug suspects, sometimes shoplifters, sometimes registered sex offenders.
More than 470,000 people passed under the cameras. Officers made 173 arrests. Only one false alert resulted in a stop—and when officers realized the error, they let the person walk. That’s the official version, anyway.
The Met highlighted one success story: a registered sex offender being arrested for communicating with a child under 16, later sentenced to two years in May after breaching his sexual harm prevention order and creating indecent images. That’s the kind of concrete outcome officials point to when pushing back against civil liberties complaints.
But it’s not just about the raw numbers. What matters more is the pattern:
- The Croydon setup was explicitly static, meaning the cameras were affixed to infrastructure rather than mobile vans
- Each deployment had a defined start-and-end timeline, and watchlists were purged within 24 hours
- Officers on the ground had real-time access to the facial recognition output and could verify matches before stopping anyone
That last point matters, because in January 2026, a different LFR operation run by Thames Valley Police led to the wrongful arrest of Alvi Choudhury, a Southampton resident who’d never even set foot in Milton Keynes. Thames Valley’s retrospective LFR system flagged him as a suspect because of racial bias in the algorithm—his photo matched someone else’s in the database, and since he’s Asian, he was the wrong kind of match for someone from an ethnic minority group. He got held for ten hours before being released with no charges.
That’s why Big Brother Watch is fighting so hard. They lost their High Court challenge in April 2026, but their argument still holds water: if the tech misidentifies people based on race—and it does—then scaling it up city-wide could normalize constant surveillance, especially for communities already over-policed.
Commissioner Rowley’s tech-enabled policing vision
MPS Commissioner Mark Rowley made his intent very clear on June 24th during the West End rollout announcement. His strategy hinges on significantly stepping up our use of technology to fundamentally change how we protect the public. That’s not a throwaway line. It’s a directive.
Rowley laid out three pillars for this tech-first approach:
- Live LFR as we know it—fixed cameras, real-time matching, ground officers intercepting flagged individuals
- City-wide emergency services drone network to extend aerial surveillance capabilities across Greater London
- AI analysis of one million CCTV cameras, crunching footage to detect anomalies, predict crime hotspots, and route incidents faster
This isn’t happening in a vacuum. The Met’s technology budget remains constrained—around £6,000 per person, less than half what some other government agencies spend. Rowley’s team is pushing back against budget cuts even as they’re asked to do more with fewer resources.
Earlier this month, the commissioner said the force would have to cut about 700 frontline posts after Deputy Mayor for Policing and Crime Kaya Comer-Schwartz refused to approve the Met’s plan to award a major contract to Palantir. The ethics of partnering with controversial data analytics firms aside, that rejection highlights just how tight things are getting. So when Rowley talks about tech-enabled policing as a force multiplier, he’s not just ticking boxes. He’s trying to maintain public safety without the manpower he says he needs.
Here’s where it gets complicated: if facial recognition were perfectly accurate—if it never misidentified someone, especially from minority groups—the case for scaling would be stronger. But we know it doesn’t work that way. The AI underpinning these systems inherits the biases of its training data, and the result is predictable: false positivesclustered around demographic groups already subject to disproportionate scrutiny.
Policing by consent vs. permanent surveillance: the ethical tightrope
Policing by consent—not by coercion—is a foundational principle in British law enforcement. It means communities trust the police, report crimes, provide witness statements, and generally assume officers are on their side. When that trust erodes, people stop cooperating. They stop calling 999 even when their homes are being broken into. They walk away from scenes instead of offering testimony.
Big Brother Watch isn’t wrong to warn that static LFR deployments risk eroding this trust. Jack Coulson, the group’s head of advocacy, put it bluntly in a press release: "We are calling on the Met to stop this experiment until, at least, Parliament has spoken." His point is simple: no democratic society should let an unelected agency install permanent biometric surveillance infrastructure without explicit legislative backing.
Parliament hasn’t spoken yet. There’s no dedicated law governing live facial recognition use in public spaces. No independent oversight body reviewing watchlists or audit trails. No standardized redress protocol when false matches lead to wrongful stops—and we already saw what happened in Milton Keynes.
Commissioner Rowley may say the West End rollout "fundamentally changes how we protect the public," but what about protecting people from overreach? When a system flags someone based on bad data, the burden currently falls on that individual to explain why they weren’t actually guilty of whatever crime was committed nearby. They get stopped, questioned, maybe held for hours—not because they did something wrong, but because the algorithm got it wrong.
That’s not a bug; it’s a feature of how these systems are currently designed. The police get speed and scale. The public gets uncertainty, embarrassment, and in some cases, lost time and trauma.
I don’t want London cops to stop catching criminals. But I also don’t want them using tools that treat entire demographics as suspects until proven otherwise. If static LFR stays in place long-term—and the Met has clearly indicated it plans to—the least we can do is demand transparency, accountability, and real legislative guardrails before the first camera goes up in Soho.