Brady O’Rourke walks to the busiest street near his house in Crystal, Minnesota, a few times a week, carrying a roof rake. He tapes a sign to the end of it, walks to the intersection, and lifts the rake into the air. The Flock camera at the corner stops recording. He waves at passing cars. Some honk.

“I don’t want to pay to be surveilled,” O’Rourke told NPR. “My tax dollars are funding cameras at the end of my block. I don’t want to live like that. I don’t think anybody should have to live like that.”

O’Rourke is one of an estimated 25,000 people registered for DeFlock’s National Week of Action against automated license plate readers — DeFlock being the advocacy group that tracks the camera footprint — protests and education events in cities across the country. The ACLU’s Jay Stanley, twenty-plus years in the privacy fight, has called the breadth of this opposition “not like anything I’ve seen.” “We’re seeing opposition from the left,” Stanley told NPR. “We’re seeing opposition from MAGA Republicans. We’re seeing opposition from mainstream Republicans, from centrists. And so it really is a nonpartisan issue.”

Garrett Langley built the cameras O’Rourke is fighting. Langley is the chief executive of Flock Safety, the Atlanta-based company whose license plate readers, by the company’s own count, are now used by more than 5,000 law enforcement agencies. Together with competitors, the cameras number more than 130,000 on American streets. Every month they capture billions of license-plate reads — along with make, model, color, and visible alterations of every vehicle that passes. The result is a searchable record of where every driver has been: to work, to the doctor, to the daycare, to the clinic, to the meeting that nobody else was supposed to know about.

Langley told NPR the cameras help police find stolen vehicles and locate missing people. A preliminary study suggests they can deter car theft. The Minnesota Chiefs of Police Association says they are “very helpful tools.”

That is the motte. The bailey is what the technology is architected to do.

The cameras do not operate the way the motte describes. They capture every vehicle, not stolen ones. The data is not stored the way the motte implies. It flows into centralized cloud networks that police departments, with Flock’s blessing, can query from anywhere — and that Flock itself can access. And the data does not stay local. It has already been shared, in documented cases, with federal agencies that local governments never authorized to see it. A sheriff’s office in Texas used it to track a woman who may have had an abortion. Other officers have used it to stalk ex-partners. ICE has accessed it for immigration enforcement without local consent.

This is the pattern the philosopher Nicholas Shackel identified in 2005 and that critics have since named the motte-and-bailey: advance the defensible claim, retreat to it under challenge, then resume the indefensible claim once the challenge has passed. The motte is “stolen-car finder.” The bailey is generalized tracking of every vehicle on every road. The architecture was never built to do what the marketing describes.

This is not a stolen-vehicle finder. It is a generalized surveillance infrastructure sold to cities as a stolen-vehicle finder.

A 2024 analysis by a California police department found Flock misread license plates in 71 percent of the alerts it sent to police. Seventy-one percent. The system is generating false alerts faster than officers can clear them, and every false alert is built on a record of an innocent driver whose movement has been logged into a permanent database that may be queried by agencies that driver has never heard of. The error rate does not just degrade the product. It converts every alert into a probable-cause claim about an innocent person. The cameras do not need to be accurate. The cameras need to be everywhere. Flock’s revenue comes from breadth — more departments, more cameras, more data, more cloud subscriptions — not from precision. A system that misreads seven out of ten plates is a system designed to sweep broadly, to capture everything, and to let the algorithm sort it later. The sorting is the product. The people caught in the sweep are not the customers. They are the inventory.

Cui bono. The benefit flows upward: to Flock Safety, to its investors, to the 5,000-plus agencies that acquire the technology without having to build it themselves, and to the federal agencies that get the spillover data without having to ask for it. The cost flows downward and outward — and it does not flow evenly. The drivers whose movements are now queryable by federal agencies are disproportionately the drivers least equipped to push back when a query comes back with their name on it. The immigrant whose every trip to work, school, and the grocery store is now in a permanent database does not have the political standing to challenge the data-sharing arrangement that put it there. The woman whose clinic visit is now part of a sheriff’s investigative file does not get a phone call before the file is opened. The architecture is built to be invisible to the people it surveils and visible only to the agencies that surveil them. That is not a bug. It is the feature that makes the architecture valuable to the people who buy it.

The asymmetry is the point. The company sells cameras to police departments. The police departments install them. The data flows to Flock’s cloud network. And the resulting apparatus — billions of plates, millions of movements, mapped and stored and searchable — is controlled by a private corporation that answers to its shareholders, not to the people whose movements it records. The company is called Flock Safety. The safety it provides is safety for its business model. The people it watches did not consent.

Langley’s response to the backlash has been the corporate version of “don’t shoot the messenger.” He told NPR that if citizens have concerns, they should take them to their city council. This is technically true in the way that the architects of a surveillance apparatus are technically not the ones doing the surveilling. Cities nominally control what data Flock shares, but the company’s architecture, defaults, and access tiers determine what sharing looks like in practice. When ICE queried Flock data, the request did not go through a city-council vote. When a Texas sheriff used the system to track a woman who may have had an abortion, the trail of evidence that fell out of the query did not pause to ask whether her city had authorized the inquiry. That is how the architecture works. It is not designed to be obeyed. It is designed to be queried. The consent it asks for is the consent of the police department that bought the contract — not the consent of the driver whose plate was captured, not the consent of the immigrant whose movements are now part of a federal database, not the consent of the woman whose clinic visit is now a permanent record that a sheriff in another state can pull up with a login and a query box.

The CEO built the messenger. The CEO decided the architecture. The CEO set the default settings that determine what cities can and cannot do with the data Flock collects. The CEO then redirected complaints to the city councils whose procurement software auto-renews the cloud subscription that funds the architecture. Revenue moves quarterly. Democratic deliberation moves on the schedule of school boards and budget meetings. The company collects on a quarterly close. The people who live under the cameras organize on a school-board calendar.

Clare Garvie, deputy director of technology law and policy at New York University’s Policing Project, puts the constitutional question plainly. The Fourth Amendment, she says, was designed to prevent exactly this: law enforcement affirmatively mapping where people go as they go about their daily lives. The cameras are already deployed across thousands of departments. The Supreme Court has not yet squarely ruled on whether generalized automated tracking of every vehicle on every road constitutes an unreasonable search. The federal appellate courts are working through the question with mixed results. The cases will take years to resolve. The cameras are being installed today. By the time the courts catch up with the architecture, the architecture will be the architecture.

Last week, Flock announced changes — shorter recommended data-retention periods, cutoffs for officers who appear to be misusing the tool. These are the kinds of changes a company makes when it has lost control of the conversation and is trying to recover it without giving up the product. None of the announced changes address the underlying structural problem. The infrastructure remains. The cloud network remains. The business model remains. The cameras remain on the poles. What changed is the window dressing: shorter retention windows that address the public objection without dismantling the apparatus that produced it. Flock Safety identified the abuse risk, built the system anyway, profited from the deployment, and when the backlash arrived, announced cosmetic adjustments while the architecture stayed intact. That is not accountability. That is a company managing its brand while the surveillance continues.

The backlash is real. According to NPR’s review of local news reports, Flock cameras have been vandalized in at least 36 states — stolen, sawed down, crashed into, covered in paint, shot at. At least 100 cities have deactivated Flock cameras or canceled their contracts. Some are switching to other vendors — to Axon, as earlier Main Street Independent reporting noted — which is the surveillance equivalent of switching brands of surveillance. Switching brands does not change the architecture. The architecture is the problem. The cameras are the given. The debate is about terms and conditions. The fundamental question — whether a private corporation should operate a mass surveillance network that maps the movements of millions of people who did not consent, and whether that corporation should control the data it collects from public streets with public money — that question is not on the table. When a city cancels Flock, Axon arrives. The apparatus is not dismantled — it is re-badged. The surveillance continues under a different logo, and the democratic energy spent canceling one contract subsidizes the next.

The apparatus of control is built in the quarter the contract is signed. Governance follows on the schedule of the governed. And by the time governance catches up, the apparatus has become the default from which the governed must negotiate their way back toward the country the Fourth Amendment imagined.

This is not a left problem or a right problem. It is the problem the Fourth Amendment was written to address and the problem the American political system has not yet figured out how to address in the age of generalized surveillance. The ACLU and mainstream Republicans and MAGA Republicans and the centrist privacy advocates are all looking at the same architecture and naming the same problem. They are correct. The question is whether the American political system has the capacity to do anything about it before the architecture becomes permanent. At the current rate of installation and the current architecture of data-sharing, it will.

The arc bends only when it is pushed. The pushing is in the city council meeting where the contract comes up for renewal. The pushing is in the courthouse where the Fourth Amendment case gets filed. The pushing is in the statehouse where the data-retention law gets written. Brady O’Rourke and his roof rake are not the whole of that. The 25,000 people registered with DeFlock for this week’s actions are not the whole of that. The 100-plus cities that have already pulled the cameras are not the whole of that. They are the beginning of that. The architecture is in place; the political coalition against it is now forming; the question is whether it forms in time to do anything.

The horizon is a country where the infrastructure of daily life answers to the people who live under it. We name what has been built. We name who built it. We name what they did. And we keep the receipts.