How many license plate readers are there?

There is no federal registry of automated license plate readers, no licensing regime that produces a public count, and no requirement that a police department disclose how many it operates. A camera can be installed by a city, a county, a private business, or a homeowners association, and in most states none of them has to tell anyone.

So every number in circulation is an estimate produced by somebody choosing a method. The useful question is not “what is the number” but “whose count is this, and what did it count?”

When national coverage quotes a figure, it is usually the volunteer map. Fox News, reporting in July 2026, put the total at about 119,000 and was explicit about where that came from: “the open-source DeFlock map displayed about 119,000 readers across the United States. DeFlock is a community-powered project that maps reported camera locations using public submissions and data from OpenStreetMap.”³

That matters for reading this site too, because it is the same lineage. Our count comes from the same OpenStreetMap tagging that DeFlock feeds and draws from. When you see a headline number and our number in the same week, you are not looking at two independent measurements converging — you are looking at one dataset, quoted twice.

We would rather say that plainly than let the agreement look like corroboration. What the volunteer count is genuinely good for is location: it says where cameras have been seen, one node at a time, with a person who saw them.

As of August 14, 2026, this site indexes 130,300 documented reader locations, refreshed every six hours from OpenStreetMap. That is not a claim about how many cameras exist. It is a count of how many have been mapped by someone, and it moves for two different reasons that look identical in the data: a camera goes up, or a volunteer finally records one that has been there for a year.

How the locations are assigned to states and cities, and what we exclude, is in our methodology.

Because OpenStreetMap keeps its full edit history, the growth of the record can be reconstructed month by month. This is that curve: every node tagged as a license plate reader in the contiguous United States, from 2019 to 2026.

The shape is the story. In January 2019 there were 30 mapped readers in the whole country. Five years later, in January 2024, there were 1,145 — real growth, but still a rounding error against what was already on the street. Then the record went vertical: 7,618 in January 2025, 63,605 a year later, roughly 8× in twelve months, and 110,433 by 2026-07.

It is tempting to read that line as cameras appearing. It is not, and the distinction is the single most important thing on this page.

A camera counts from the day it was mapped, not the day it was installed. Much of the 2025 rise is volunteers catching up with hardware that had been standing for months or years. The curve is a measure of attention as much as of infrastructure.

Organized mapping moves it. A coordinated push in one metro area produces a step in the national line that has nothing to do with new installations there.

The series has edges. It counts nodes inside a bounding box around the contiguous states, so Alaska and Hawaii fall outside it, and it will differ slightly from the live total above, which is computed differently and on a different day.

Removals are quiet. When a city cancels a contract and takes cameras down — and a number have — the map only reflects it once someone edits the nodes.

Read the curve as what it is: the best available record of how quickly this infrastructure became visible to the public. The Electronic Frontier Foundation’s work on ALPRs is a good companion for what the technology does once installed.⁴

The national figure is the least interesting thing about this data. Counts by state and city, with vendor breakdowns and per-square-mile density, are in the camera index, and every documented location is on the map.

The underlying locations are © OpenStreetMap contributors and available under the Open Database License;⁵ the history series is computed with the ohsome API over the full OSM edit history.¹²