Skip to main content

Politicians make promises on their stump — we watch and hold them accountable.

Help keep the record honest →Create an account

Issue

Mass Surveillance, Tracking, and Flock Safety

Camera and sensor networks that record where ordinary people go without any suspicion of wrongdoing: automated licence-plate readers, led by Flock Safety, together with facial recognition, biometric identification, cell-site simulators, phone-location data bought from brokers, and aerial and drone surveillance. Tracked here as a constitutional question before a public-safety one — who authorised the cameras, how long the record of your movements is kept, which agencies can search it, and whether a warrant is ever required.

In short

Automated plate readers photograph a state-mandated identifier on a public road, and under current law that is generally not treated as a search. Knotts holds there is no expectation of privacy in public vehicular movement, though Knotts itself reserved the question of dragnet practices. The one federal appeals court to decide the question found no search, reasoning from a town system of no more than ten cameras. No federal appeals court has yet ruled on a national commercial network. StumpWatch's position is different, and we state it as a position rather than as law. A company running a surveillance network on the government's behalf is acting as an agent of the government. The test is whether the same act would violate the Constitution if an officer carried it out directly. If it would, contracting the work to a private firm does not cure it. These were never only plate readers. The vendor's own policy records the vehicle, its characteristics, the time and the camera's location, and a competitor's patent claims reach tyre-pressure sensors expressly. There is no way to opt out: a driver is not a customer and was offered no terms. The documented failures are administrative rather than constitutional, and they are real. Agencies enabled federal access without knowing it, and one Texas query reached more than 83,000 cameras, a figure the Illinois Secretary of State has now stated in an official record rather than the press. Opinion accepts this surveillance in some settings and rejects it in others, and these records cannot show a trend of their own, because they capture almost every question at a single point — though the pollsters’ own reports sometimes carry earlier readings not ingested here. The nearest thing to a repeated item, concern about how government uses collected data, ran 64 percent in 2019 and 71 percent in 2023, but Pew changed the wording between the two, so it is not a clean like-for-like comparison. The page also measures the vendor's public statements against the documentary record, and reports where the two diverge without asserting anyone's motive. Most of those items remain unresolved, and are labelled as such. Four hundred and seventy-five officials are on record here across 761 positions. Most of that is voting record: 581 are roll-call votes, and 517 of those concern FISA Section 702 rather than plate readers. The remaining 180 come from oversight letters and public statements, and all 180 are from Democratic members — we have not located letters or statements on this issue from Republican members, and we do not know whether that reflects the record or a gap in our collection. Even a constitutional win may produce no remedy. The constraints that bite today are state statutes with audits, purge deadlines and penalties.

Explore the record

Everything behind this page sits in the 7 sections below — the claims, who is on record, what the public thinks, and every citation. Choose one.

Constitutional analysis

This is our assessment of the issue against the Constitution — not a sourced fact and not a court holding. It separates what the law establishes today, what is genuinely contested now, and the argument for where the law should go. Written and adversarially reviewed away from this site, then stored; the page renders it and never generates it. Everything behind the tabs below is that assessment, it is open to dispute, and the counter-arguments that survived review are published in full under Summary.

Does government use of automated licence-plate readers violate the Constitution, and are vendors like Flock Safety acting as agents of the government?

Key points

  • Almost every enforceable limit on plate-reader networks in the United States today comes from state legislatures, not from the Fourth Amendment.
  • No federal appellate court has held that querying a plate-reader system is a search; the one decision that reached the question reasoned from a municipal system of no more than ten cameras.
  • StumpWatch's position is that a company running a surveillance network on the government's behalf is a government agent, so the test is whether the same act would violate the Constitution if an official did it directly.
  • Courts have not adopted that reasoning on these facts, and this page says so rather than presenting the position as settled law.
  • These devices were never only plate readers: the vendor's own policy records the vehicle, its characteristics, the time and the camera's location, and a competitor holds a patent whose claim 27 covers tire pressure sensors and other vehicle systems by name.
  • There is no evidence the largest vendor collects Bluetooth, Wi-Fi, cellular or tyre-sensor signals, and the Minnesota audits often cited against that claim never actually examined the question.
  • There is no way to opt out. A driver is not a customer, was offered no terms, and the vendor states it cannot process data-subject requests for plates in its footage.
  • Even winning the constitutional argument may produce nothing: good faith, qualified immunity and standing each defeat a remedy independently of the merits.
  • The documented failures so far are administrative rather than constitutional, and they are real, including federal access that agencies did not know they had switched on.

What is at stake

WHAT IS AT STAKE

A camera on a pole photographs your license plate as you drive past. By itself that is nothing: a plate is a number the state requires you to display so strangers and police can read it. But thousands of those cameras, wired together, retained for weeks, and searchable after the fact, produce something different in kind, a record of where a particular car has been. Whether the Constitution has anything to say about that is the question, and the honest answer today is: much less than most people assume. Almost every enforceable limit on these systems in the United States right now comes from state legislatures, not from the Fourth Amendment. This page holds two things. First, that a dense, retained, retrospectively searchable plate-reader network is the kind of surveillance the Fourth Amendment should reach. Second, and more fundamentally, that a private company running such a network on the government's behalf is acting as an agent of the government, so the constitutional question is not changed by routing the work through a contractor. Courts have so far largely rejected the first and have set a demanding bar for the second. An earlier draft of this page presented both as settled doctrine. They are not settled, and that was the error worth correcting — not the positions themselves, which we hold, and which we set out and defend below. What follows separates what courts have actually decided, what is genuinely open, and what we hold to be true regardless of how the litigation turns out.

Counter-arguments that survived review

These were argued against the analysis at full strength and were not defeated. They are published here, unfolded, because an assessment that hides its strongest surviving objection is advocacy rather than accountability.

Five independent reviewers were asked to refute this analysis rather than improve it. Across forty load-bearing claims they returned 171 verdicts of "survives" and 29 of "overstated". No claim was refuted outright by a majority. These are the objections that survived, and readers should weigh them. THE KNOWING-EXPOSURE RULE IS THE HARDER OBSTACLE, NOT THE THIRD-PARTY DOCTRINE. The strongest government answer to a plate-reader claim is not Smith v. Maryland but New York v. Class: the exterior of a car is thrust into the public eye, and it is unreasonable to expect privacy in an object the law requires be displayed in plain view. That rule is older than the third-party doctrine and was left untouched by Carpenter. If it controls, no amount of aggregation helps, because nothing private was ever observed. This objection reframes the whole question and the analysis now addresses it directly. THE AGGREGATION THRESHOLD MAY NOT EXIST. The analysis treats "enough cameras, held long enough" as a line that will eventually be crossed. A reviewer argued there may be no such line. The two rationales behind the only federal appellate holding on point, that a plate carries no privacy interest and that a roadside sensor never follows anyone indoors, do not change at ten cameras or ten thousand. The contrary authority is dictum; the authority against is a holding. VIRGINIA'S STATUTE DOES NOT ENLIST THE VENDOR. An earlier draft cited Va. Code 2.2-5517 as evidence that a state had put a vendor in a searching role. Restoring the omitted words reverses it: the vendor may access system data only "for maintenance and quality assurance purposes". Virginia confined the vendor to being a maintenance technician. That claim was withdrawn rather than repaired. SKINNER'S PERMISSIVE HALF WAS QUOTED WITHOUT ITS LIMITS, and both real-world agency examples offered in support produced no relief for anyone. The narrowed agency argument survives only where the government directs the specific act complained of, which is a much smaller target than the original thesis. THERE MAY BE NO REMEDY EVEN IF THE CONSTITUTIONAL ARGUMENT WINS. The first draft never mentioned the exclusionary rule, good faith, qualified immunity, standing or mootness. A reviewer pointed out that the 2026 decision the analysis treats as its best news is the proof against it: a court found the warrant plainly unconstitutional and admitted the evidence anyway on good faith. This produced the remedy section now in the analysis. TWO CITATION DEFECTS WERE CAUGHT BEFORE PUBLICATION AND REMOVED. One cited case did not exist. One quotation attributed to a police department appears in no source. Both were struck. Neither was load-bearing, and the propositions they had been attached to either stand on other authority or were dropped. They are recorded here because a reader is entitled to know that the review found them.

How this was reviewed

Assessment · AI analysis · 86% confidence
Claims assessed
40
Survived review
38 of 40
Adversarial passes
5
Confidence
86%

Version 10 · published 2026-08-10 · every earlier version is retained, never deleted.

StumpWatch is live, and the record is still growing. Many promises and positions aren’t tracked yet, and some features are still in beta. Add a sourced promise and help keep the record honest.

Help keep the record honest →