Is it legal for police to use live facial recognition technology?

Thank you for listening to this edition of The Law Show, the podcast that helps you understand the law. In this week's show, we're looking at the law enforcement technology that may be coming to a street near you. It's called Live Facial Recognition, a kind of supercharged CCTV for the police, which uses AI to recognize faces in a crowd from live video. The Home Secretary has announced a rollout of the technology to all English and Welsh police forces, and she rejects the idea that it's like big brother watching you. But the legal framework allowing police to use it is a mishmash of different laws, and it's being challenged in court. We'll be looking at the laws in surveillance and speaking to a BBC correspondent who's seen live facial recognition in action. And speaking of technology, the law on flying a drone has changed, and there are hefty fines for breaking the rules. I'll be speaking to a lawyer who specialises in drone law later in the podcast. But first, the Home Secretary Shabana Mahmoud says that she makes no apology for announcing the rollout of live facial recognition to all police forces in England and Wales. Police say it's a vital tool in the fight against crime, but civil liberties groups say it's authoritarian and a step towards a surveillance state. Now you might be familiar with facial recognition cameras being used in shops. The difference here is that the software used by police tracks faces against a watchlist from a live video feed. Under a government white paper on policing, the number of live facial recognition vans will increase from 10 to 50. 13 police forces have already used the technology, but the biggest pilot was held by the Metropolitan Police in London. BBC London's Home Affairs Correspondence, Sonja Jessup, has seen the cameras in action. Sonja, can you describe to us exactly how this works? I saw a couple of different deployments with Met that I was reporting on. One of them was in East London using a van. This is one of a number of regular deployments we understand that are taking place across London. And it involved a red van with facial recognition signage on the side of it. Inside the van, a couple of members of the team who are monitoring the live footage coming from the cameras. They're looking at faces of people walking past. We saw how, as people pass the camera, their face perhaps came into vision and then would be pixelated. If that's someone who's not on the watch list, not someone that they're looking for, then their data is immediately impermanently deleted. So they were saying you're watching that in real time. Now outside, there were cameras that were on the roof of the van. They were beside the van. They were a little further down into the marketplace based on a temporary pole that had been set up. And there were officers on the ground who have handheld devices. If someone walks past who the technology recognises as someone on that list, the handheld devices will beep, alert the officers on the ground. They will then have to identify who that is and go over. And then they are supposed to check if that person is the person that they are looking for. In another example, I saw was in Croydon, which is a pilot scheme that they're running. And they have no vans in place. Instead, the cameras are on street furniture. In this case, it was street lamps, no vans, the team are monitoring remotely, we're told. But the officers are still on the ground with their handheld devices. The mettax explained that the benefits to them is that they don't have a sort of set up process. They say it's more efficient to use it that way. They put up signs around the area. I have to say I had to go and look for the signage. I did find it in both cases. I wanted if I was being a bit unobservant. Some of the people I spoke to passing by, I made a point of asking them when I was asking them how they felt about this technology. I asked them, did you see the signage? Were you aware this is an area? Everyone was surprised. Some of them had read about a life-ish recognition in their area. But they weren't aware that that was an area they were stepping into. Many of them were not concerned about it. In fact, many of them were supportive of it. Some people were concerned about it and they were surprised that it was going on. So what sort of places have we seen that's been used? In the first instance, I mentioned with the van, this is one of the frequent deployments we see. That was in Waltham Stowe, in East London, in the marketplace. They said that the kind of deployments they're choosing are based around hotspot crimes. They told me that that particular area was a hotspot for mobile phone theft and robbery. The watchlist, they describe it as a bespoke watchlist, which is one that's put together that morning, which contain tens of thousands of wanted offenders. They say, "People who are wanted." But it doesn't. I said, "So are all they specifically about mobile phone theft and robbery?" But they were saying, "No, it could be all sorts of crimes." The hotspot area may not necessarily relate specifically to the people on that watchlist is my understanding. They also, as I say, have used it at large events such as the King's Coronation. They've used it at an all-thlander football derby, for example, between Tottenham and Arsenal on the approaches to those matches. They've also used it more controversially at the Notting Hill Carnival last year. Now, that provoked some concerns, some anti-racism and privacy campaigners, wrote the commissioner to express concern. They said they feared that this technology could be less accurate for people of colour, and they questioned why the carnival had been chosen. Now, the commissioner acknowledged that in previous years, back in 2016, 2017, there had been a use of technology which had flagged 100 false positive alerts and no arrests had been made. So he acknowledged that that technology used then had not been started. - We should have spoken about false positive. - Yeah, false positive. This is when someone is flagged up. The technology says, "We think this is a person that's wanted." But then when it's checked out, when that person is stopped, it turns out that that's not the wanted suspect. And in this particular case, back in 2016 and '17, this had been an issue. The commissioner acknowledged that. He said since then the technology has improved. The algorithm, the way we're using it has changed. So we're confident in the way we're using it. But there was a lot of concern over it. And again, I mentioned the Croydon pilot scheme. Campaigners were questioning to me, "Well, why has Croydon been chosen?" Croydon has a higher demographic population of black people, for example. Why have you chosen that specific area? So, in fact, do you interview a man who had been misidentified by the technology? - Yeah, and this is a man called Sean Thompson. He was stopped a couple of years ago outside London Bridge Station. And he was told that he was a wanted suspect. And he said that he was held for around 30 minutes. He was asked to provide his fingerprints. And he refused to give his fingerprints. And he said he was eventually released after providing a photograph of, I think, it was his passport photo. Now, he himself is black. He said it felt to him like it was stop and search on steroids. And he felt it was very intimidating. Now, I asked the Met Police about his specific case. Now, the Met wouldn't comment specifically because he is bringing a high court challenge alongside the director of the Privacy Campaign Group Big Brotherwatch Silky Carlo. And there's been a two-day hearing at the high court because their argument is that the Met is using this technology unlawfully. The Met disputes this as it is using it lawfully. It's confident in the way it's using it. We haven't yet had a decision handed down in that case. - How successful have the pilots been in identifying serious offenders? - Well, the Met would say it's been very successfully refers to it as groundbreaking technology. So between January 2024 and January 2026, for example, they told me that they made more than 1,700 arrests of what they described as dangerous offenders, for offenses including rape and child abuse. They also flagged up as being a way of checking on people such as registered sex offenders. In one example, they gave, they had stopped a man, he was a registered sex offender to check, up on whether he was keeping to his sexual harm prevention order, which banned him from being in contact with children. And he was in the company of a six-year-old girl. He was also in possession of a knife, and he was arrested and jailed for two years. Now, the Met said, without this technology, of course, yes, we would be keeping checks on people and checking up on them, but they said we were able to sort of identify him actually in the company of a child and the mother knew nothing about it. So they felt that was a powerful example of how that's being used. And of course, they campaign as question whether this is the right way to go about it. - Sonia, thank you. Although there is a legal basis for the police to use facial recognition technology for law enforcement, this relies on common law powers and a patchwork of the qualities, human rights and data protection laws, as well as police guidance. Those are not our words, but the words of a Home Office public consultation which ended in February. And the legality of live facial recognition has been challenged in the High Court by Sean Thompson's legal team. We're still waiting to hear the result of that judicial review. So what are the laws covering the use of this technology? Let's speak to Professor Karen Young, who's at the Law School at Birmingham University, and Dr. Astroskikay, senior lecturer in AI, disruptive innovation and law at Brunelle University of London. Astros, we talk a lot about reasonable suspicion to detain someone, but is a pinging on the piece of technology enough for reasonable suspicion, Astros. No, not really. The reasonable suspicion for committing an offense has to be established based on evidence and robust police intelligence. We know that technology can misidentify people that has been proven, and just because someone is wrong. happen to be identified by the technology, police cannot establish reasonable suspicion on that basis alone. But if police believe that the person that the technology has flagged is the right person and the person is reasonably suspected to have committed a crime, then of course they can be detained. So Karen, these are deployed in public areas. Do the police need legal permission to put these cameras up in those areas? They cannot use this technology unless they can demonstrate that they have a legal basis for using it and that's why there are ongoing disputes about whether or not that legality has been properly established. So let's explain that. Why do we mean that there might be an issue with regard to the legal framework for using these? There are a patchwork of laws that in fact do apply to this technology. So it's not accurate to say that there's no law. There is law, but it's really complicated and a big chunk of it upon which police are currently relying lies in the common law. So it's not. Let's explain what the common law is. Yes, that's a good question. So typically when we think about laws made by our democratically elected body, we're thinking about laws that have been promulgated by parliament. They have been publicly debated. They are published so that we can reflect on and put our responses to MPs and select committees very scrutiny before they pass into law and then they are placed on a statute book which are openly available. And so if we want to know what the law is, we can look it up on the internet and find out what the law is. The common law on the other hand is judge made law and this is part of the law of the land, but it is acquired through decision making through case law. And that means that people like me, lawyers have to extract what the relevant legal principles are from this big volume of case law that grows over time to try and work out what the law actually is, what are the boundaries of the law, what is acceptable and what is unacceptable. So what principles are we relying on to tell police how they can use this technology? Yeah, so in fact we do have one case, the bridges case, which actually went from the high court to the court of appeal. So it decided that there was a basis in the common law for allowing the technology, but in order to satisfy the legal test for legitimately violating people's privacy, but it also decided every time that camera takes a picture of you and passes your image against that watch list, your privacy is interfered with. And that means it becomes necessary for the user of the technology, the police to justify the privacy intrusion. And it needs to be necessary and proportionate in a democratic society. And in the bridges case, there was too much discretion and therefore the use was unlawful and my understanding is this is the same kind of argument upon which Sean Thompson is currently seeking to overturn the use of a facial recognition for which he was misidentified. Asruss, is this an issue? Is the fact that we don't have a very clearly go framework, such an issue for the use of the technology? The current laws are not sufficient to define clear boundaries for when, where and how the technology can be used. So it is not disputable that the lack of clear legal framework is the problem. But at the same time, we need to recognize that the existing legal framework does allow police to use facial recognition technology. So the common law principle, which is invoked by the police, is based on the idea that police have brought power to gather evidence in the course of investigating crimes and safeguarding public safety. And in its 2020 decision in bridge the high court acknowledged that these common law broad power of police can be extended to facial recognition technology as well. In public, do we just have an expectation of the right to privacy? So the legal test applied by courts here in the UK, as well as be by the European Court of Human Rights in Strasbourg, is whether the processing of personal data, whether that interferes with people's reasonable expectation on privacy. And so the court said, given the fact that it's a sensitive personal data, which is being extracted from people's faces in public spaces where they are going about their day. And that biometric data is being processed and compared with personal data on a watch list. It does interfere with the privacy rights. Karen, a lot of this has centered around the watch list. So do you have any concerns about it? Yeah. How long have you got? The police say that they are only uploading to the watch list. The faces of people wanted for serious crimes. What does that mean? Well, indeed, it's a good question because if you actually look at the watch list, the size of them has expanded quite rapidly over time. In the London deployments, they started with only a couple of hundred people. Now they are putting the faces of tens of thousands of people on these watch lists. They're putting images of people who wanted for traffic offenses, for minor drug offenses, and for shoplifting. I really don't think that justifies putting your face on a watch list and stopping you in the middle of the street. When we say shoplifting, it has variations. Some shoplifting crimes are committed by organized gang members and they also involve violence against retail workers and stealing of valuable items. And if we put a blanket restriction on what kind of crimes can justify the use of facial recognition technology, this could end up undermining public security. So I think whatever approach that we take has to be more nuanced and has to give a room for responsible use of the technology to address crimes that might even appear to be minor when looked at individually, but can become a great public threat when they are combined or committed in certain ways. Asters, given the historical problems of AI recognizing black and brown faces and those of women, is there still a problem with racial and gender bias in the system? The men, police's own facial recognition technology have been tested by the National Physical Laboratory and they have found that the technology disproportionate misidentifies women of color. But that also depends on the so-called face-match threshold. Essentially this is a setting which determines how accurately the technology identifies, people's faces, and at higher setting it identifies less people but more accurately. But at lower threshold it identifies more people but with more inaccuracy. And the men, police currently use the setting which was recommended by the National Physical Laboratory and that setting reduces ethnic disparity to so-called statistically insignificant level. Do you think that the current framework is fit for purpose, Karen? Does it actually, on the evidence in real-world settings, contribute to more effective prevention of crime and detection of those who are wanted for serious offenses, for that we have little demonstrable evidence. So, don't we need to deploy the technology in order to get the evidence? Yes, so exactly, we need to conduct legally responsible and epically sound real-world testing. But what would legal and ethical look like in practice if faces are being deleted upon recording? So for a start you need to be really clear about where you're going to deploy it and why. You need to clearly justify which images will be placed on the watch list. The source of those images must be legally acquired and then when you deploy it, you need to have proper training of your offices. You need to have a clear protocol for what happens when there is a match alert, recognizing that the mere generation of a match alert is not reasonable grounds for suspicion. So you need to understand your intervention with the public is on the basis of consent and that people are not obliged to tend at identification evidence. If you do not have reasonable suspicion, you need to be properly recording how often you have false negatives and false positives. As a risk coming to you. I think the technology has certainly been beneficial. If we look at the Met Police's annual report for 2025 in a period of one year, about 900 people have been arrested and quarter of these arrests related to violence against women and girls. The 2025 Notting Hill Carnival was the safest in many years and there was reduction in violent crimes like robbery down by 70% compared to the previous year. I think the evidence shows that the technology is indeed beneficial and as you said, if we don't use the technology, how could we know whether it's useful or not? Testing needs to be within a responsible legal framework that also builds public trust. Clearly we need a legal framework that creates boundaries for what kind of crimes the technology can't be used and when it can't be deployed and who can authorise it and at the moment that really doesn't exist most of the stuff is done by police force themselves. Just a reminder that this rollout is planned for England and Wales. In Scotland, no decision has been made regarding the implementation of live facial recognition. Police Scotland is exploring the potential use of live facial recognition in consultation with the public and stakeholders. The police service of Northern Ireland has sent us a statement saying that it does not currently use live facial recognition technology and has no existing capability, contracts or funding in place to deploy it. The PSNI goes on to say that no decision has been taken to introduce the technology at present. Professor Karinyoung, Dr. Asreski Kai and Sonja Jessup, thank you. Now, there seems to be no end to the uses of drone technology. Most recently, they transformed coverage of the Winter Olympics, drone displays that complement fireworks are now a thing, and Amazon has just started testing delivery flights at their centre in Darlington, in County Durham. There is also a darker side to the technology. Drones armed with explosives and anti-personnel weapons have become a permanent fixture in modern wars in both Ukraine and the Middle East. But they're not just for broadcasters, businesses or familiar use. It's estimated that nearly 400,000 consumer drones were sold in the UK last year, and domestic drones can be incredibly cheap. For example, my producer has bought one for me with a 4K camera that cost 1999. I cannot guarantee the quality of this drone, or even if it has a full 4K camera, some of the user reviews are quite skeptical, but I can guarantee that if I go outside and start flying it right now, I'll be breaking the law which changed in January. Richard Ryan is a barrister from Blackestons, specialising in drone and unmanned aviation law. Richard, can you tell us about how the law has changed? The law changed in January, as you rightly said, which captures now drones weighing 100 grams or more, and with a camera. So now we have drones that have to be registered, and you'd have to register with a flyer ID and an operator ID. The flyer ID is an online theory test that you'd have to take with a civil aviation authority, which is about 40 multiple choice questions based on the drone code, and the operator ID is something that you'd have to register for annually at a cost of just under £12 a year to show that you're the responsible person that owns that drone. Much like a vehicle registration is on the car, you would get an identifying number that you would place and attach to your drone. So what kind of questions would be on the online flyer ID test? To start with, you need to know the limitations of flying a drone, and there are boundaries in place. The theory test will test your knowledge on those boundaries. For example, you can only fly it to a certain height. You can only fly it away from uninvolved people. There may be privacy concerns, and also some good safety practices that will be in the questions. So it sounds quite sci-fi, but what is the drone code? So drone code is literally what it says on the tin. The civil aviation authority have published an easy to use and understandable framework for pilots that just buy a drone off the shelf and want to go and fly it. It's got infographics on it, so it's easily readable and digestible. Hopefully, it educates the drone user into operating the drone safely. What are the penalties for not having one of these flyer IDs? So there could be fines that could be imposed upon you as a drone operator, or not having a flyer ID, because ultimately this is a registration system, so that drone can be identified to you as a responsible operator of that drone. So we were able to get this drone in about a day for a very reasonable price, not that I'm advertising, yes. But we didn't need to give a flyer ID or prove anything to get it. Is this a problem with enforcement? Enforcement's always a challenge, as it is with any regulation. And I know that some years ago, the civil aviation authority were putting in drone code leaflets with every drone that was purchased. Now, I'm not sure if you got one of those when you purchased your drone. Nope. Oh, okay. Well, that could be an issue. But ultimately, you can buy a drone without having to show the shop or the online presence that you've got a flyer ID, because that comes after in the sequence of purchase and flyer ID. But where are we allowed to fly? Definitely not in the studio, but where else? Even though they're very small drones, and drones are classified under aviation law as an aircraft, what the civil aviation authority is trying to do here is mitigate risk of damage to persons, vehicles, vessels and structures, which is fairly old terminology. But where can you fly an answer to your question? You can't fly in restricted zones, such as airports, air strips, space ports even, and also one thing to be very wary of, especially with small drones, you might think you can go to your local park. Quite often, local councils have bylaws in place that restrict takeoff and landing. The other areas where you can't in terms of flight restriction zones, or FRZs, as they're commonly known, prisons and military sites and even nuclear power stations. So in these areas where you're not allowed to fly, could they shoot your drone down? Well certainly that police and law enforcement do have powers to interdict, or whether they decide to shoot or not is another question. Because as we've seen with Ukraine and also more recently in the Middle East, the proliferation of weaponized drones has increased significantly. So the question is you don't always know the intent of the drone operator when flying a drone, so if you don't know the intent, how can you quantify that risk? If someone finds a drone over my property and I think it's an invasion of my privacy, what can I do? This is becoming a very common question, because you want to protect your privacy and you don't want to be intruded upon by a drone and they can be quite noisy at times. And what we're seeing here is a real tension between airspace law and land law. Now airspace above your property is generally speaking uncontrolled airspace, much like a helicopter can fly over it today with no issues. So can a drone. It's not illegal to fly a drone over your property. It starts to become a gray area or a legal if I'm being intrusive and I use the camera and I do something with those recorded images. Could it be legal for me to shoot down a drone if it's close enough to me to do so? No, no. Because the drone itself is classed as an aircraft under UK law. In fact there was a case in America where somebody did shoot down the drone with their shotgun. So that has happened already. But I would certainly not suggest you do that because you could find yourself in all kinds of trouble. Okay, what if I didn't use a gun? What if I used like a catapult or a tennis ball or like a very highly trained hawk? They still have to be careful. Simply because under I think it's section 76 of the Civil Aviation Act, there's a strict liability. So if you did take action that took a drone out of the sky and it did fall in the injure somebody, then there will be consequences for that. Richard Ryan and two all my other guests today, thank you. That's all for this week. But as always, if you have a question about the law, you can contact us at the law show at bbc.co.uk and our whatsapp number is o3306784700. We don't give legal advice, but we can help explain how the law works in future podcasts. The law show was presented by me, Dr. Joal Grogon and produced by Ravi Nike and Charlotte Rules. The editor was Tom Bigwood. If you enjoyed this program about the law, you can subscribe to learn more.