About

Mass surveillance has always had a limiting factor, and that factor was people.

Governments have watched their own populations for centuries, and every one of those efforts hit the same problem. A spy can only report what they see and hear. A wiretap only matters if somebody listens to what it records. A camera only matters if somebody watches the tapes. To follow a whole population you need an enormous number of watchers, and then you need people to watch the watchers. Most of what has ever been collected has never been reviewed by anyone, and the reason was volume and time rather than law.

Artificial intelligence (AI) removes that limiting factor. Surveillance analysis is no longer constrained by how many people are dedicated to it. It is constrained by how much computing power is available, and computing power is expanding quickly. AI can review every plate, every recording, and every image an agency holds, without a human ever needing to look at any of it.

A warrant is a judge's permission to search, granted only when police can point to specific facts suggesting a crime.¹ It is slow on purpose. Passive surveillance does not ask for a warrant, because it is not looking for anyone in particular. An automated license plate reader, meaning a roadside camera that photographs every plate that passes it and logs the time and place, keeps that record whether or not anyone is suspected of anything, and an officer can search the stored record later.² Courts have so far treated that as outside the Fourth Amendment.

Why local

Mass surveillance does not arrive all at once. It arrives as line items: a grant-funded camera in one budget cycle and a body camera refresh in the next, a school safety package, a records system upgrade. Each purchase is approved on its own terms, and on its own terms each one is reasonable. Nobody votes on the whole system, because the whole system is never the agenda item. The result is not a surveillance plan, it is an accumulation.

A county is where that sum becomes countable again. We are starting with rural counties because that is where the whole picture is still small enough for one person to hold. Every device can be counted, every contract read, every meeting attended.

For each county we publish what we can document. Which systems are operating, who runs them, what they cost, what the approving body was told, and what the public record does not answer.

Our standards

Every fact on this site names its source and the date we last checked it. Agency-confirmed records and community-reported sightings are labeled separately and never combined in a single count, because the distance between the two is part of what the map shows. The methodology page spells this out in full.

Two things you can do

Look at your own county. Start with what is documented where you live, then take it to the people who approved it. Agendas, minutes, and contracts are public, and a resident who has read them changes the room. See the counties we cover.

Reduce your own exposure. Knowing which systems operate near you, what they capture, and how long they keep it tells you where your own choices still matter. See what is documented in your county.

Footnotes

1. Congressional Research Service, IF13068, July 21, 2025 (a warrantless search is "presumptively unreasonable"; probable cause and a neutral magistrate are required). ↩
2. Congressional Research Service, IF13068, July 21, 2025 (description of automated license plate reader operation, and that federal courts have generally found no reasonable expectation of privacy in this data as used so far). The underlying holdings are United States v. Knotts, 460 U.S. 276, 281 (1983) (no reasonable expectation of privacy in movements on public thoroughfares) and New York v. Class, 475 U.S. 106, 114 (1986) (no privacy interest in a plate the state requires be displayed). ↩