Last Saturday, May sixteenth, London saw simultaneous rival protests that turned into what the Met called one of the largest policing operations of the year. Protesters are saying the facial recognition deployment was unprecedented, and there was a heavy plainclothes presence throughout. So Daniel wrote in asking something that cuts right to the mechanics of it — how does plainclothes policing actually work in practice? Is it a dedicated unit, or do regular officers just swap their uniforms for street clothes? How do they maintain cover when the crowd is watching for exactly that? And do unmarked police vehicles rotate license plates between operations, or is that just a spy-movie detail?
I love this question because it lands right in the gap between what people imagine and what's actually happening on the ground. And last week gave us a perfect stress test — you had two groups that actively despise each other, plus a heavy police footprint, plus facial recognition vans at entry points. If you wanted to see plainclothes tactics under a microscope, that was the moment.
Before we dig into last Saturday, let's pin down what we're even talking about. These get collapsed into one thing in the public imagination, and they're really not.
A plainclothes officer is a fully sworn, rank-and-file police officer who happens to be wearing civilian clothes for a specific operation. They're not living a false identity. They showed up to work, got a briefing, changed into a hoodie and jeans, and went out. They carry their warrant card, they're identifiable to other officers through coded radio identifiers, and they go home at the end of the shift. An undercover officer — what the Met would call a Special Operations officer, from the SO10 directorate — is a different animal entirely. Those are officers with long-term cover identities, fake backstories, the works. They're deployed for months or years under a completely different legal and procedural framework.
The difference is basically: one guy clocked in and changed his shirt, the other guy has a fake lease agreement and a fabricated employment history.
In most UK forces, plainclothes officers aren't a permanent, separate unit. They're drawn from Territorial Support Groups or local Basic Command Units. You might be on a response team one week in full uniform, and the next week you're tapped for a plainclothes shift at a protest because intelligence says there's a risk of violence and they need eyes in the crowd.
Which raises the first thing I think a lot of people get wrong. The image is of some shadowy dedicated secret-police division. The reality is more like: Dave from the public order unit got told to leave his helmet at the station today.
That's both reassuring and slightly alarming. Reassuring because it's not a permanent secret police force. Alarming because the selection and training is minimal. There's no formal qualification for plainclothes duty in most forces. Officers volunteer or are selected based on local intelligence needs. They get a briefing on the day's objectives — here's who we're monitoring, here's what we expect, here's where you position — and a clear prohibition on engaging unless absolutely necessary. Then they go out.
No formal qualification. So what happens when Dave from the public order unit, who's used to the authority that comes with a uniform, suddenly has to blend in and not act like a police officer?
This is exactly the tension the College of Policing flagged in their twenty twenty-five review. They found that plainclothes officers reduced physical confrontations by twenty-three percent in high-risk protest settings, which is a genuine operational win. But they also noted that the lack of standardized training creates what they politely called "risks of mission creep and procedural errors." Translation: officers without proper training sometimes default to behaviors that blow their cover or escalate situations they should be de-escalating.
Twenty-three percent reduction in confrontations is a real number though. That's not nothing.
It's not. And the mechanism makes sense. If you're a protester and you see a line of uniformed officers with helmets and shields, your behavior changes — and not always toward calm. But if the person next to you looks like another protester, you're less likely to perform for them. The mere possibility that the person next to you might be police changes the risk calculus for people considering violence.
How does this actually work on the ground? Walk me through the mechanics of a plainclothes team at something like last Saturday's protests.
Let's take a typical deployment. You've got teams of four to six plainclothes officers. One is designated as the spotter — that's the person whose job is purely to observe and communicate. They're not making arrests, they're not engaging. They're on a covert earpiece to a control room, feeding information back. The spotter identifies persons of interest, tracks movements, and coordinates with uniformed units positioned out of sight but close enough to move in.
The spotter is the director. Everyone else is waiting for cues.
The other officers are positioned at different points in the crowd. They maintain separation from each other — you don't want four people who all look like they're scanning the crowd in the same way standing together. They use what's called "natural cover" — standing near a food vendor, looking at a phone, having a conversation with each other that's actually coded communication.
The earpieces — are these visible? Because that seems like a quick way to get made.
They're typically in-ear monitors that sit deep in the ear canal. Very hard to spot unless you're inches away. The radio itself is concealed under clothing. But you're right that this is a vulnerability — there have been cases where protesters have spotted the subtle tell of someone talking without moving their lips, or the slight bulge of a radio pack under a jacket.
How do they avoid getting burned? There's that famous footage from the Just Stop Oil protests in twenty twenty-three where protesters literally pointed out plainclothes officers they recognized from previous operations.
That was a tactical disaster for the Met, and it led to a full review. What happened was that the same officers had been deployed to multiple Just Stop Oil actions over several months. Protesters — who are often highly organized and share information on encrypted channels — built face databases. They'd photograph officers at one protest, share the images, and when the same officer appeared at the next action, they'd be identified and surrounded. The protesters would chant "we know who you are" and film them, which effectively neutralizes the officer's utility for that operation.
The countermeasure is... don't send the same face twice?
But it's harder than it sounds. The pool of officers willing and available for plainclothes duty isn't infinite. And some officers develop expertise — they get good at reading protest dynamics, they know the tactics, they're effective. The operational instinct is to use your best people. But the counterintelligence reality is that using your best people repeatedly burns them. The twenty twenty-three review pushed for stricter rotation policies and shorter deployment windows for individual officers.
What about the rendezvous problem? If a plainclothes officer needs to talk to a uniformed colleague, they can't exactly walk up to the police line and have a chat.
This is one of the most operationally delicate parts of the whole thing. A plainclothes officer being seen talking to uniformed officers is an instant burn. So they use pre-arranged rendezvous points — inside a police van with blacked-out windows, behind a building, inside a nearby shop that's been quietly co-opted for the operation. They also use dead drops for equipment — a bag left in a specific location that another officer picks up, no direct contact required.
Like something out of a Cold War spy novel.
The parallels are not accidental. A lot of these tactics were developed by people who studied intelligence tradecraft. The King's Cross operation in twenty twenty-four is a perfect example. Plainclothes teams used a rented delivery van as a mobile observation post — completely unmarked, civilian plates, parked in a loading zone. Officers would cycle in and out through the back doors, which faced away from the protest route. The protesters had no idea the van was a police asset until after the operation.
Which brings us to vehicles. Daniel specifically asked about unmarked cars and license plate rotation, and I have to admit I've wondered about this too. Do police actually swap plates between operations?
This is one of those questions where the truth is both more boring and more clever than the fiction. No, UK police do not routinely swap license plates on unmarked vehicles between operations. What they do instead is maintain a very large pool. The Met's unmarked fleet is approximately two hundred cars — typically VW Passats, Skoda Octavias, BMW three series, nothing flashy. For any given operation, they pull from this pool. So the car you see at one protest has a completely different plate from the car you saw last week not because they changed the plates but because it's a different car.
Plate tracking by protesters becomes impractical not because plates change but because the fleet is large enough that you'd need to catalog hundreds of vehicles to reliably spot repeats.
Protesters do try. There are online databases where activists log unmarked police vehicles — make, model, color, plate number, location seen. But with two hundred cars rotating through, and with vehicles being retired and added regularly, maintaining a useful database is genuinely difficult. The Met will only swap plates if a specific vehicle is compromised — meaning it's been photographed and widely shared on social media in a way that makes it unusable for future operations.
What's "widely shared" in operational terms?
There's no fixed threshold, but the general practice is that if a vehicle's plate appears in a public forum with its police role identified, and that post gets significant circulation, the vehicle gets pulled from the rotation and re-plated. The old plates are surrendered to the DVLA. It's a reactive measure, not a proactive one.
The protesters who think they're tracking police vehicles are playing a shell game where the shells keep getting replaced.
And the police know the tracking efforts exist, which is why they've also moved toward using entirely civilian-looking vehicles for some operations — rental vans, as we saw at King's Cross, or cars seized in criminal investigations that haven't been re-registered yet.
Now let's talk about the facial recognition piece, because that's what made last Saturday's operation novel. Not the technology itself — the Met's been using Live Facial Recognition since twenty twenty — but the coordination between LFR and plainclothes teams.
Let me explain how the LFR system actually works in a protest context, because the term "facial recognition" gets thrown around like it's one thing and it's really a specific operational pipeline. The Met deploys LFR vans — marked vehicles with cameras mounted on top, typically positioned at entry and exit points to a protest area. The cameras feed into a system that compares faces against a watchlist of up to two thousand individuals. These are people with outstanding warrants, people subject to court orders, people identified as posing a specific risk. It's not scanning against the entire population — it's a curated list.
Two thousand names. That's a surprisingly small list for a city of nine million.
It is, and that's by design — partly legal, partly practical. The system needs to return matches quickly enough to be operationally useful, and a smaller watchlist means fewer false positives. When the system flags a match, a human officer reviews it. That's the "human in the loop" verification step that the High Court emphasized in the twenty twenty-three Liberty ruling. If the human reviewer confirms the match, the information goes to officers on the ground.
This is where the plainclothes teams come in.
At last Saturday's operation, plainclothes officers were positioned near the LFR vans at entry points. When the system flagged a match and a human reviewer confirmed it, the plainclothes team would move to intercept the individual. The idea is that a plainclothes approach is less likely to trigger a confrontation or a crowd reaction than a uniformed officer wading in to make an arrest.
The LFR is the net, and the plainclothes officers are the hands.
That's the model. And what made last Saturday unprecedented wasn't the facial recognition — LFR has been used at protests since twenty twenty-two. It was the scale of the simultaneous deployment. You had two rival protest groups, multiple entry points, LFR vans at each one, and plainclothes teams positioned at every van. The coordination required was significant.
The protesters called it unprecedented, and in one sense they're right — the coordination was new. But the technology wasn't.
Right, and this is where the public conversation often gets muddled. People hear "unprecedented facial recognition" and imagine the police deployed some new, more powerful system. They didn't. The watchlist was the same size it always is. The cameras were the same ones they've been using. What changed was the operational design — embedding plainclothes officers directly into the LFR pipeline in a way that made the whole system faster and less visible.
Let's talk about the legal framework for a minute, because plainclothes officers at protests are operating under specific statutory powers that a lot of people don't know about.
Section sixty of the Criminal Justice and Public Order Act nineteen ninety-four. This allows police to designate an area where officers can stop and search without reasonable suspicion. Normally, a stop and search requires an officer to have grounds — specific intelligence or observed behavior that justifies the search. Under a Section sixty authorization, that requirement is suspended. An officer can search anyone in the designated area.
Plainclothes officers can conduct these stops?
They can, but there's a crucial rule: if a plainclothes officer is conducting a stop and search, they must wear a visible identification armband. This is supposed to ensure that the person being stopped knows they're dealing with a police officer and not a random stranger demanding to search them.
Which makes sense as a safeguard. But I recall that armband requirement got... flexible during some of the twenty twenty-four operations.
It did, and this was controversial. During several protest operations in twenty twenty-four, the Met allowed plainclothes officers to delay displaying their armbands until after they'd made initial contact and assessed the situation. The reasoning was officer safety — if you're approaching someone you believe may be violent, broadcasting that you're police before you're in control of the situation could be dangerous. The counterargument, which civil liberties groups made forcefully, is that a stop and search without immediate identification is functionally indistinguishable from an assault.
Where did that land legally?
It's still contested. The Met's position is that the armband rule allows for operational discretion. Liberty and other groups argue that any delay in identification violates the spirit of the safeguards the High Court demanded in the twenty twenty-three ruling. There hasn't been a definitive court decision on this specific point yet.
We've got a legal gray zone sitting at the intersection of officer safety and civil liberties. That's not going to resolve itself cleanly.
It never does. And this is the broader tension that runs through all of plainclothes protest policing. On one hand, the College of Policing review found that these tactics reduce violence. Twenty-three percent fewer physical confrontations represents real people not getting hurt. On the other hand, the chilling effect on lawful assembly is real and hard to measure. If you believe there might be plainclothes officers in the crowd, if you know there are facial recognition cameras at the entrance, does that change your willingness to attend a protest? And if it does, is that a feature or a bug?
The Home Office seems to think it's a feature. I saw they're piloting something called "covert facial recognition" in Birmingham this year.
Yes, the twenty twenty-six Birmingham pilot. This is the next evolution, and it's worth paying attention to. The idea is to move facial recognition from fixed vans at entry points to portable devices — handheld units or body-worn cameras with LFR capability. If that works, the distinction between uniformed and plainclothes policing starts to dissolve. A plainclothes officer could walk through a crowd, and their body-worn camera — which looks like any other piece of tech — is running faces against a watchlist in real time.
The surveillance layer becomes completely invisible. No vans, no marked cameras, no visible infrastructure at all.
That's the trajectory. And the Home Office's framing is interesting — they're pitching it as an efficiency measure. Why position vans at entry points when officers can do the same work while moving through the crowd? But the civil liberties implications are significant. If facial recognition becomes portable and covert, the only way to know you're being scanned is if the police tell you — and the Met's current practice of publishing LFR deployment notices twenty-four hours in advance might not survive the transition to portable systems.
The deployment notices are an interesting transparency measure. I didn't realize those were public.
They are, and they're one of the concrete outcomes of the Liberty ruling. The High Court found that LFR was lawful but required the Met to publish notices of where and when LFR would be deployed. You can find these on the Met's website — they tell you the date, the location, and the general purpose of the deployment. It's not perfect transparency — the notices are broad, and they don't tell you who's on the watchlist — but it's more than most people realize exists.
If you're planning to attend a protest, you can actually check whether LFR will be in use.
And that's one of those practical things that doesn't make headlines but matters for people who want to understand what they're walking into.
Let's pull on another thread Daniel raised. The challenge of maintaining an effective covert presence when the protest is "on guard" — when protesters are actively looking for plainclothes officers. How do the police adapt when the crowd knows the playbook?
This is where the tactical creativity comes in, and it's a constant cat-and-mouse game. When protesters are actively hunting for plainclothes officers — and some groups have dedicated "cop-watchers" whose entire role is to identify and film police — the standard playbook stops working. You can't just put four guys in hoodies in the middle of the crowd and expect them to blend in.
Because the cop-watchers are looking for exactly that. Four people who don't quite fit, who are scanning rather than participating, who drift toward the edges of confrontations.
The police adapt. One adaptation is embedding officers in specific sub-groups. If the protest has a medical tent, you put an officer there who actually knows first aid. If there's a legal observer contingent, you embed someone who can talk the talk. The cover becomes more sophisticated — you're not just wearing civilian clothes, you're playing a role that gives you a reason to be moving through the crowd and observing.
The medical-tent officer is actually providing medical assistance while also gathering intelligence.
That's the ethical knot at the center of all this. That officer might help someone who's injured. The intelligence gathering doesn't negate the medical assistance. But the person receiving treatment doesn't know the person treating them is police, and that raises questions about consent and deception that don't have easy answers.
Another adaptation I've read about is the use of female officers in plainclothes roles, specifically because the stereotype of a plainclothes officer is male.
That's a significant operational shift over the last decade. The public image of a plainclothes officer — and to be fair, the historical reality — was overwhelmingly male. Protesters looking for police were looking for men. Deploying female officers in plainclothes roles exploits that cognitive bias. A woman with a phone, standing near the front of a crowd, is far less likely to be identified as police than a man doing the same thing. The Met has been deliberately diversifying its plainclothes teams for exactly this reason.
The protesters' cop-watching is only as good as their stereotypes.
The police know exactly what those stereotypes are. They train against them.
Let's come back to the vehicles question for a moment, because there's another layer to this. Daniel asked about high-stakes operations specifically. Do police even use unmarked vehicles in those contexts, or do they rely entirely on other methods?
They do use unmarked vehicles, but they're increasingly cautious about how and where. In a high-stakes operation with rival protest groups — like last Saturday — the risk calculation changes. An unmarked police car parked near a protest route is a liability if it's identified. Protesters have been known to surround and block unmarked vehicles, which not only neutralizes the asset but creates an incident that can escalate quickly.
What's the alternative?
Both, plus fixed observation posts. The Met has access to the National Police Air Service helicopter for aerial surveillance, and they've been expanding their drone capability for protest monitoring. A drone with a good camera can provide crowd intelligence without any officers on the ground being exposed. But drones have their own limitations — battery life, weather, airspace restrictions in central London. So the unmarked vehicle remains a key tool, just one that's deployed more carefully than it used to be.
There's also the question of what happens when an unmarked vehicle is compromised mid-operation. You've got officers in a car that's just been identified and photographed. What's the protocol?
The vehicle is pulled from the operation, the officers are reassigned or stood down, and the car goes back to the fleet pool for re-plating. The intelligence value of the vehicle is gone, and the officers in it are potentially at risk if the crowd becomes hostile. The Met's priority in that scenario is getting people out safely, not salvaging the operation.
The operational cost of a compromised vehicle is high enough that they're incentivized to be careful about vehicle deployment.
This is why the rental-van tactic from King's Cross was clever. A rented delivery van with civilian plates, parked in a loading zone, is functionally invisible. Even if protesters notice it, they have no reason to connect it to the police unless they see officers entering or exiting. And the Met was careful about that — officers cycled in and out through rear doors facing away from the protest route, and they did it in ones and twos, not in groups.
It's the banality that provides the cover. A delivery van is the visual equivalent of beige wallpaper. Nobody looks twice.
That's exactly the principle. The most effective covert vehicle isn't the one that looks like a spy car — it's the one that looks like it belongs in the environment so completely that it becomes invisible.
Let's shift to something that I think gets lost in all the tactical discussion. The human toll on the officers themselves. You're a retired doctor — what does rotating between uniformed and plainclothes duty do to someone psychologically?
And there's not enough research on it. But anecdotally, the transition is harder than it looks. A uniformed officer has a clear identity. The uniform signals authority, it provides a script for interactions, it offers a degree of physical protection — both practically and psychologically. When you take the uniform off, you lose all of that. You're alone in a crowd that may be hostile to police, you can't reveal who you are, and you're making decisions that could have serious consequences without the immediate backup of colleagues.
You're doing this after what sounds like a fairly minimal briefing.
The briefing covers operational objectives and rules of engagement. It doesn't cover the psychological experience of standing in a crowd of people who, if they knew who you were, might turn on you. Some officers find it exhilarating — it's closer to detective work than public-order policing. Others find it stressful in ways that accumulate over time.
The ones who find it stressful — do they have a way to opt out?
In theory, yes. Plainclothes duty is voluntary in most forces. In practice, there's professional pressure. Saying no to a plainclothes assignment can be seen as lacking commitment or being difficult. The College of Policing review touched on this briefly — they recommended clearer opt-out procedures and psychological support for officers after plainclothes deployments, but it's not clear how widely those recommendations have been implemented.
We've got a system where officers are temporarily reassigned to a role with minimal training, operating in a psychologically demanding environment, with legal powers that are still being contested in court, and the public doesn't fully understand what they're seeing. That's a lot of pressure points.
Last Saturday compressed all of those pressure points into a single afternoon. You had rival protests, which means the risk of violence was higher than a typical demonstration. You had facial recognition vans at multiple entry points. You had plainclothes teams embedded throughout. And you had protesters who were actively looking for police presence and filming everything. If something had gone wrong — a misidentification, an escalation, an officer whose cover was blown in a dangerous position — it could have gone very wrong very fast.
The fact that it didn't is either a testament to the professionalism of the officers involved or luck. Probably some of both.
But "it didn't go wrong this time" isn't the same as "the system works." The College of Policing's twenty-three percent reduction in confrontations is a real finding. But the lack of standardized training, the legal gray zones around identification during stops, the psychological burden on officers — these are not resolved issues. They're just ones that haven't produced a catastrophe yet.
What does this mean for someone who's not a police officer and not a protester but who cares about how policing works in a democratic society?
I think there are three things worth holding onto. First, plainclothes policing is not a shadowy secret police force — it's a tactical rotation of regular officers. That's both reassuring and concerning. It means there's no permanent plainclothes unit developing its own institutional culture outside public oversight. But it also means the officers doing this work may not have the specialized training the role demands.
The transparency is structural, not procedural. The structure prevents a secret police from forming, but the procedures don't guarantee that every plainclothes interaction is handled well.
That's well put. Second, the combination of plainclothes officers and facial recognition creates a surveillance layer that is new. Not because either component is new, but because the integration makes both more effective. A plainclothes officer without facial recognition is just eyes in a crowd. Facial recognition without plainclothes officers requires uniformed intervention that can escalate situations. Together, they create a system that can identify, track, and intercept with minimal visible police presence.
Third, there are concrete things people can do if they're attending a protest and want to understand what they're walking into. Check the Met's website for LFR deployment notices — they're published twenty-four hours before operations. Know that plainclothes officers are likely present but are bound by rules of engagement — they must identify themselves if conducting a stop and search, even if there's some operational flexibility about when. And filming interactions is legal and can serve as a deterrent against misconduct.
The phone camera as accountability mechanism. It's not a perfect tool, but it's the one people have.
It's been effective. Some of the twenty twenty-three tactical reforms happened specifically because protesters filmed plainclothes officers who'd been deployed too many times to the same group. The footage created pressure that led to policy changes. That's not nothing.
There's one question that's been nagging at me through this whole conversation. If the Birmingham pilot succeeds — if portable, covert facial recognition becomes operational — does the distinction between uniformed and plainclothes policing even matter anymore? Every officer becomes a walking surveillance node. The uniform becomes almost decorative.
That's the question, isn't it. The Home Office is clearly moving in that direction. The Birmingham pilot is explicitly about testing whether LFR can work on body-worn cameras and handheld devices. If it can, then the operational logic pushes toward universal deployment. Why have dedicated LFR vans at entry points when every officer can do the same work continuously?
At that point, the entire framework we've been discussing — plainclothes versus uniformed, covert versus overt — starts to collapse. You're not being watched by a van you can see or an officer you've identified. You're being watched by the crowd itself, or more precisely, by devices in the crowd that look like phones and body cameras.
The Met would argue that this is just efficiency — doing with portable devices what they currently do with vans. But the qualitative difference matters. A visible LFR van at a protest entrance is a form of notice. You know you're being scanned when you walk past it. A plainclothes officer with a body-worn camera running facial recognition in real time — you don't know. There's no notice, no transparency, no opportunity to make an informed choice about whether to be in that space.
The deployment notices become meaningless. "LFR will be deployed in the Whitehall area" — that's not informative if every officer in Whitehall is carrying it.
The notice model assumes fixed infrastructure. Portable LFR breaks that assumption.
The open question we're left with is whether the legal framework can adapt faster than the technology. The Liberty ruling required transparency measures that made sense for van-based LFR. If the Birmingham pilot succeeds, those measures are obsolete before the ink is dry.
Parliament isn't exactly known for moving quickly on tech regulation. The Investigatory Powers Act took years. The Online Safety Act took years. Facial recognition is evolving on a timeline of months.
We've covered a lot of ground. The mechanics of plainclothes teams, the vehicle fleet shell game, the LFR integration, the training gaps, the legal gray zones, and where this is all heading. If someone wants to go deeper on the surveillance law side, there's an episode we did on the legal frameworks for police surveillance technologies that lays out the statutory landscape in more detail.
For the protest rights angle — what the law actually says about police powers at demonstrations, including Section sixty and the limits on stop and search — that's worth understanding if you're going to be in this space.
The throughline here is that the operational reality of protest policing is more sophisticated and more constrained than the headlines suggest, but the constraints are eroding faster than the public conversation is keeping up with.
The constraint that matters most — transparency — is the one that portable technology most directly threatens. That's what I'd want listeners to sit with. Not whether plainclothes policing is good or bad in the abstract, but whether the systems we've built to make it accountable can survive the next generation of tools.
Now: Hilbert's daily fun fact.
Hilbert: In the Azores, it was once traditional for whaling crews to paint their oar blades with a mixture of whale oil and locally sourced indigo, creating a deep blue pigment that was believed to be invisible to sperm whales when viewed from below against the daytime sky.
I have so many questions about the optics of whale vision and none of them are going to get answered.
The whales saw the boats just fine, Hilbert. They just didn't care.
This has been My Weird Prompts, produced by Hilbert Flumingtop. If you enjoyed this, leave us a review wherever you get your podcasts — it helps other people find the show.
I'm Corn.
I'm Herman Poppleberry. Thanks for listening.
