Facial recognition: Met officers photograph your face, no consent

Quick answer
Can Met Police officers take your photo for facial recognition without consent?
Under the current Met pilot, yes. Around 100 officers can photograph a person with a phone app and compare the image with about one million custody photos. Officers must say why they are doing it, but they do not need permission. Asking what happens to the photo afterward is the practical step.
Picture a police officer on a London street, holding up a phone and taking your picture. You didn't say yes, and you aren't required to. That is the basic idea behind a new facial recognition pilot (a trial of software that compares a face in a photo against a stored collection of faces) being run by the Metropolitan Police.
Facial recognition from the Met Police now includes a six-month pilot where 100 officers can photograph people without consent, so the smartest move is to ask what the photo is for and how to learn what happens to it afterward.
Let's answer the 11pm question first: no, this isn't about every person in London getting scanned by a camera on a lamppost. It's narrower than that. It's also more personal, because it starts with an officer pointing a phone at one human being.
What Is This Facial Recognition Pilot, in Plain English?
According to the reporting gathered for this piece, 100 Met officers have been cleared to check faces using an app on Met-issued phones. The app compares the photo an officer takes against a database of around one million custody images (photos taken of people while they were held by police). Officers must tell you why they're taking your picture. They do not need your consent. The pilot runs for six months, and you can read the basics in the Yahoo News report.
One thing to be clear about: the reporting I have doesn't describe any one person caught up in this. So I'm not going to tell you a story about "a man in Croydon." This is about the rule, and who it could touch.
100
officers cleared to check faces from a phone, against about one million custody images, in a six month pilot
Source: reporting on the Met Police pilot, via Yahoo News UK This article is part of a series, start with Fargo Facial Face Match Jails Tennessee Woman 108 Days.
Operator-initiated facial recognition: a face check an officer chooses to run
The technical name for this is operator-initiated facial recognition. Translation: a person (the "operator") decides to run the check. That's different from the cameras you may have read about, which scan crowds automatically and look for faces on a watch list (a list of people police want to find). Here, the officer picks the moment and the person.
Is that better or worse? Honestly, both. A human choosing means it's more targeted. But a human choosing also means a human can choose badly, or unfairly. Hold that thought.
How recognition software reads a digital image of your face
Facial recognition software starts with a digital image, which is a grid of pixels. It finds a human face in the captured image, measures facial features such as the spacing of the eyes, and gives a facial similarity score against stored facial images. Most facial recognition systems work this way, and a high score points toward the same face identity without proving it. Because biometric data like a face cannot be changed the way a password can, privacy and security questions follow any collection of it.
Why Met Police Facial Recognition Without Consent Feels So Different
We're used to the idea that a camera on a street might catch us. Most of us have made a quiet peace with that. A phone held up at your face is another thing entirely. It's close, it's deliberate, and it's about you.
Here's a trick your brain plays, by the way. You've probably seen stories about facial recognition mistakes, so they feel common and close to home. Psychologists call this the availability heuristic (judging how likely something is by how easily examples pop into your head). So it's worth keeping your head here. This is a small pilot, 100 officers, six months. But small pilots are exactly where the rules for the big rollout get written.
The pilot is putting "powerful surveillance technology in officers' pockets" and "combines all of the problems of stop-and-search with those of facial recognition."
Liberty's advocacy director, as reported by Great Driffield Radio
Liberty is a UK human rights group, and that is a sharp way to put it. Stop-and-search (when police stop you on the street and check you) has long raised questions about who gets picked. Add a face check to that, and you stack one worry on top of another.
StopWatch UK raised a related concern. Its executive director said not enough was being done to "mitigate inherent racial bias in facial recognition technology." Bias, in plain terms, means the software or the people using it may treat some groups worse than others. The same group said the technology would not make ethnic minorities "feel safer."
What does the law actually say about police facial recognition in the UK?
Not much, and that's the strange part. According to the research I was given, the UK has no law written specifically for police use of facial recognition. Police have worked from a patchwork of guidance for over a decade. The Home Office (the government department in charge of policing) has since published a response about a legal framework (a set of official rules) for how law enforcement can use biometrics, which means body data like your face. So the trial is happening while the rulebook is still being written. Previously in this series: How To Bypass Age Verification 1 In 3 Teens Already Know.
Facial Recognition Cameras vs. a Phone in an Officer's Hand
These two things get lumped together, and they shouldn't be. Here's the difference, based on what the reporting describes.
| Question | Crowd cameras (live facial recognition) | Officer phone check (this pilot) |
|---|---|---|
| Who starts it? | The system scans faces as people pass | An officer chooses to photograph one person |
| What is the face compared to? | A watch list | About one million custody images |
| Is consent needed? | Not asked of passers-by | No, but the officer must say why |
| Legal status | No specific UK law | No specific UK law |
Notice the last row. Different tools, same missing rulebook. That's the part I'd underline twice.
Why Met Police Facial Recognition Matters Beyond London
- Consent is optional here: officers must explain themselves, but your permission isn't part of the deal.
- Pilots set habits: whatever 100 officers do in six months becomes the starting point for the next rollout.
- The rules are late: the test is running before a dedicated law for police facial recognition exists.
What Should You Ask If an Officer Takes Your Photo?
You can't stop the photo, going by what's reported. But you can ask. Officers are required to tell you why they're taking it, so start there. Then ask the questions the reporting leaves open: where does the photo go, how long is it kept, and how can I find out what happened to it afterward?
I can't promise you'll get neat answers on the spot. I don't know, from what's been published, whether a photo that finds no match is kept or deleted. That's precisely why it's worth asking. Write down the date, place and officer's number if you can. Boring? Yes. Useful later? Very.
Face matching gives a clue, not proof of identity
If you've ever wondered whether a photo or profile is really who it claims to be, that's the exact question this kind of tech exists to answer. Done carefully, facial comparison (lining up two photos to see if they likely show the same person) can be a good tool. The difference is who controls it, who checks it, and who you can ask when it gets something wrong. A match score (a number for how alike two faces are) is a clue, never a verdict.
Met Police facial recognition now lets officers photograph people without consent, so the one thing within your control is asking why the photo is taken and how to find out where it ends up.
Recap: A six month London pilot lets 100 officers check faces from a phone without asking permission. Ask what the photo is for and how to follow it afterward. Up next: Facial Recognition 100 Met Officers Snap Faces No Consent.
Here's what sticks with me. Officers have to tell you why they're photographing you, but nobody has to ask whether they may. Will anyone be able to tell you, six months from now, what happened to your face?
facial recognition: Frequently Asked Questions
Can police take my photo without my consent in London?
In this pilot, yes. Reporting says 100 Met officers can photograph people for a face check, and they must explain why but do not need consent. The pilot lasts six months and uses phones issued by the force. The photo is compared with around one million custody images. If it happens to you, ask what the photo is for and how you can learn what happens to it afterward.
What is operator-initiated facial recognition?
It means a person, usually an officer, decides to run a face check, rather than a camera scanning everyone automatically. The officer photographs one person with a phone app, and the software compares that photo with a stored collection of faces. It is more targeted than crowd scanning, but it still depends on human judgment, and critics worry that judgment can be unfair.
Is there a specific law for Met Police facial recognition?
According to the research behind this article, no law in the UK specifically covers police use of facial recognition. Police have relied on a patchwork of guidance for over a decade. The Home Office has published a response about a legal framework for law enforcement use of biometrics, which are body details like your face, so rules may be coming, but the pilot is running before they arrive.
What is a match score in facial comparison?
A match score is a number showing how alike two faces look to the software. A higher number means a closer likeness, but it is not proof. Software can be wrong, and bias can affect results, which is what groups like StopWatch UK worry about. Treat any match as a lead that a human must check, never as a final answer about who someone is.
Why do groups like Liberty object to police facial recognition?
Liberty's advocacy director said the pilot puts powerful surveillance technology in officers' pockets and combines the problems of stop-and-search with those of facial recognition. The concern is that checks may fall hardest on some communities. StopWatch UK also said not enough is being done to reduce racial bias, and that the technology would not make ethnic minorities feel safer.
What are custody images and why do they matter?
Custody images are photos police take of people held in custody. In this pilot, the officer's phone photo is compared against about one million of them. They matter because the quality and fairness of any match depends on what is in that collection. Being photographed in custody is not the same as being guilty, so who is in the database is a fair question.
Facial recognition is a branch of computer vision, which is the field of making software understand pictures. In a facial recognition check, the software first runs face detection to find where a face sits in the image. Then it uses facial landmarks, such as the corners of the eyes, the nose tip and the edges of the mouth, to line the face up before it does anything else.
Next, the software can create a face embedding. A face embedding is a long string of numbers that stands for the pattern of one face. Face recognition then means comparing that face embedding with the stored embeddings of other faces and asking which ones sit closest. The result is a similarity score, and as noted above, a score is a lead and not a verdict.
Most modern face recognition technology relies on machine learning (software that learns patterns from many example photos instead of following hand-written rules). That matters for fairness. If the example photos the software learned from lean toward some groups, the facial recognition results can be less accurate for others, which is the kind of racial bias StopWatch UK raised. It is a fair question to put to anyone who runs a facial recognition tool.
Face recognition and face detection are not the same thing. Face detection only answers "is there a face in this image, and where?" Facial recognition goes a step further and asks "whose face is it, or which stored faces look like it?" The extra step is where most of the privacy worry comes from, because it links a face to a possible name or record.
Some people also mix up facial recognition with identity verification. Identity verification is when you choose to prove who you are, for example by showing an ID and a selfie to open an account. You know it is happening and you take part. A street photo taken by an officer is different, because the person in the picture is not the one starting the check.
You may also hear about liveness detection. Liveness detection is a test used in some identity verification apps to check that a real, live person is in front of the camera and not a printed photo or a screen. Face liveness checks of this kind are about stopping fakes. Nothing in the reporting on the Met pilot says whether any face liveness step is part of it, so I won't guess.
Privacy is the heart of the debate. A face is not like a password, because you carry it everywhere and cannot change it. When any police force collects a face image, people reasonably want to know how long it is kept, who can see it, and whether a photo with no match is deleted. Those privacy questions are why asking the officer matters.
Think about what facial recognition can and cannot do. It can analyze one image and rank stored faces by how alike they look. It cannot read intent, it cannot prove guilt, and it cannot tell you why someone was in a place. That is why a human must check any result, and why the training and judgment of the officer holding the phone count so much.
It also helps to know that live crowd systems are designed to track faces as people walk past a camera, while the phone pilot looks at one person at a time. Both rely on the same basic recognition steps: find the face, measure it, and compare it. The difference is who starts the check, which is why the table above separates the two. Neither one has a law written just for it.
If you want a simple rule for judging any facial recognition story, use three questions. What image went in, and how good was it? What was it compared against, and who chose that collection? And what happens to the image afterward? A blurry photo, a skewed database or a missing deletion rule can each weaken a face recognition result, so each question is worth asking.
Finally, remember that recognition is a clue and not a conclusion. A good facial recognition process keeps a human in charge, records what was done, and lets people ask questions about their own image. The Met pilot will be judged on exactly those points. Until clear rules exist, asking calmly why your face was photographed is the practical step open to you.
