What a plate read becomes

A single plate read records a car at a place and time. A database of reads records movements. This page follows one read from a camera to a nationwide search, a data-broker profile, and a live video feed, from court records, audit logs, and APD’s briefings. When the Electronic Frontier Foundation obtained 2.5 billion license-plate scans from 200 law-enforcement agencies, 99.5% of the plates scanned belonged to people suspected of nothing. (EFF Data Driven dataset) Retained and pooled, those scans of everyone become searchable history about anyone.

This page walks through what that means for you, concretely, using court opinions, leaked audit logs, a U.S. Senate investigation, and Asheville Police Department’s own statements to council.

A plate scan is a location record

The highest court of Massachusetts said it plainly in Commonwealth v. McCarthy (2020): aggregated plate-reader data can produce

“a highly detailed profile, not simply of where we go, but by easy inference, of our associations — political, religious, amicable and amorous.”

The court held that “individuals have a reasonable expectation of privacy in the whole of their physical movements”, and that widespread ALPR networks “could implicate constitutional protections against unreasonable searches.” (Full opinion · Harvard Law Review analysis)

How fast does the diary fill? In Norfolk, Virginia (about 175 Flock cameras, a network smaller than what Asheville plugs into), one plaintiff’s car was photographed 475 times in four and a half months. A federal judge upheld that system in January 2026, but went out of his way to warn that the answer was contingent: “at least in Norfolk, Virginia, the answer is: not today”, and that “as the number and capabilities of ALPR cameras expand, the constitutional balancing could conceivably tip the other way.” The Institute for Justice is appealing. (WHRO · Courthouse News)

Your regular routes tell anyone with database access where you sleep, where you work, which church or clinic or meeting you attend, and whose driveway your car spends the night in. The routes are the record.

The network your plate joins

Here is Deputy Chief (now Interim Chief) Jackie Stepp, briefing council on March 7, 2025, a briefing that only happened because a councilmember read about the cameras in the news and asked (video, 1:14:41):

“It’s important also to note, through our Flock Safety and Axon partnerships, that we have access to over 177,000 cameras across the Eastern US — and we’ve had that access… this includes law enforcement agencies, which makes up about 97% of that number… but it also includes business partners… we have a number of Home Depot, Lowe’s, various retail that have joined in.”

Note the tense: we’ve had that access, before council was ever told. In the same briefing she described the reach as “cameras within a 500-mile radius of where we are.” Four months later, APD told the Public Safety Committee the network was “connected to over 19,000 cameras across a 500 mile radius.” The two figures have not been reconciled on the record.C., Atlanta, Nashville, and the ocean.

And the sharing runs both ways. Every plate scanned in Asheville becomes searchable by the network, which, per audit logs from around the country, includes agencies you have never heard of, in states you have never lived in. In Spokane County, Washington, records showed the database holding local plate scans was searched 2.3 million times in six months, with over 95% of those searches from outside agencies, including Texas and Alabama. (RANGE Media)

How a nationwide search works

A Flock network search is account-based: any logged-in officer at any participating agency can query every camera shared with the national network in one step. The system requires one thing before running the search: a “reason,” typed into a free-text box by the officer, checked by no one.

One documented search: In May 2025, a sheriff’s deputy in Johnson County, Texas ran a nationwide Flock lookup for a woman who had self-managed an abortion. His search swept 6,809 networks and 83,345 cameras, including cameras in states where abortion is legal. The reason he typed into the box: “had an abortion, search for female.” (404 Media · EFF’s audit-log analysis) The full story is on Who Gets Watched.

After months of oversight correspondence with Flock, Senator Ron Wyden’s office put the structural conclusion in writing; this is a primary document, October 2025:

“Abuse of Flock cameras is inevitable, and Flock has made it clear it takes no responsibility to prevent or detect that.”

The letter quotes Flock’s own Chief Communications Officer: “it is not Flock’s job to police the police.” And it documents why the “reason” box is not the safeguard it sounds like: in EFF’s dataset of 11.4 million nationwide searches over six months, more than 14% of the reasons were just the word “investigation”: no case number, no crime. The ACLU found officers could type “hehehe” as a reason and run the search. (ACLU)

APD told 828NewsNOW in August 2026 that 18 people can search its cameras and “you have to put the reason” (828NewsNow); the Wyden letter’s findings concern the same reason field, and the count of 18 does not include outside agencies searching through the network.

From your plate to your name

Flock’s Nova product. In 2025, leaked materials showed Flock building Nova, a people-lookup tool designed, in the words of the leak, to let police “jump from LPR to person,” connecting plate scans to people-search products, public records, and data-broker material; one contemplated source was data from the 2021 ParkMobile breach. After the reporting, Flock said Nova “will not supply any data purchased from known data breaches or stolen data.” Flock’s own marketing for Nova today: search “RMS, CAD, LPR, jail records, public records, and approved open sources without switching tabs.” (404 Media · Flock’s walk-back · Flock’s product page) Nova is shipping, not hypothetical: San Diego PD quietly signed a Nova contract in April 2026 before pausing it under scrutiny. (Axios San Diego)

The sequence: a camera reads a plate; the reads become a location history; the history is joined to a name, records, and broker files. No step requires a warrant.

When the machine is wrong

Documented errors fall into three kinds:

Bad entry. In Aurora, Colorado, a typo (the numeral “1” entered for the letter “I”) sat in a state hotlist for roughly two years. In August 2020 it flagged Brittany Gilliam’s SUV as a stolen vehicle (the actual stolen vehicle: a motorcycle, from Montana). Police held Gilliam, her 6-year-old daughter, her sister, and two nieces face-down on hot asphalt at gunpoint. The city paid a $1.9 million settlement. (CNN · CBS News)

Bad read. In Española, New Mexico, a camera read a “2” as a “7”, and a 12-year-old girl ended up in handcuffs. A month later in the same town, a 17-year-old honors student was held at gunpoint after another ALPR mistake. (CBS News investigation)

Stale entry. A California privacy-board chair was held at gunpoint at Thanksgiving after his rental car was flagged stolen: the car had already been recovered; nobody had cleared the flag. (Same CBS investigation.)

A hotlist alert is built from human data entry, machine vision, and list maintenance; each has failed in the cases above.

Beyond plates: the live-video layer

The RTIC isn’t just plate readers; that’s the part Asheville already had. What the Axon contract adds is Fusus, and Axon’s marketing describes it: a CORE device placed on a private camera network “detects, analyzes and connects to every camera on the building’s network,” and the platform “extracts and unifies live video, data and sensor feeds from virtually any source,” giving police “a live, map-based view of both public and private camera networks.” (Axon’s product page)

Asheville’s version of this was described on the record at the March 25, 2025 council meeting by APD technology specialist James Wingo (video, 42:09): private participation comes in two tiers: “a registry, just so we know where a camera is in a city, and then there’s also one where you could provide us full viewing capability.” That includes, in his example, “the Ring camera at your home.” When Councilmember Roney pointed out that courts elsewhere have compelled private footage, Wingo conceded: “if the court comes to your house and has a warrant for that Ring camera footage, then you would have to give it to us.” The “voluntary” in voluntary integration belongs to the camera’s owner, never to the people it records. See Before You Connect Your Cameras for what that means if you own the camera, and Cameras in Public Housing for what it means if you live under one you never agreed to.

The sequence

A camera on your commute reads your plate, along with 99.5% of plates belonging to no suspect anywhere.
The scan joins a searchable regional and national pool: 19,000 cameras or 177,000, depending on which APD briefing you attend.
Any of thousands of network agencies (or, per the audit logs, federal agencies through them) can query it with a reason no one checks.
A hit, right or wrong, can escalate to the RTIC’s live layer: a map of public and private cameras, including homes and businesses that “opted in,” tracking in real time.
And the vendor is building the product that turns the plate into a person.

Each step is documented above from primary sources. No step requires a warrant, a judge, or a council vote; the city’s commitments on each are on the Promise Tracker.

Step 1 by itself, a camera reading the plate in front of it, checked against a hot list and then discarded under a local retention policy with an audit trail someone reads, is a different object from steps 2 through 5, and it is not what this page warns about. Every rung above it is a separate decision: to retain, to pool, to join the national lookup, to add the live layer, to buy the person-lookup product. Each of those decisions has a default set by the vendor, and each can be declined, capped, or put under a public vote. The communities in Communities that ended or limited camera contracts did exactly that.

Explorer