What is a Flock camera?

A Flock camera is an automated license plate reader — an ALPR — built by Flock Safety, an Atlanta-based company ⁷ founded in 2017 ³. It is a fixed, roadside camera that photographs vehicles as they pass and converts each photograph into a database record that can be searched later. The name has become a generic term the way “Xerox” once did: many people call any pole-mounted plate reader a Flock camera, though Motorola Solutions, Genetec, and Leonardo sell competing hardware that works on the same principles. Our visual identification guide covers how to tell the manufacturers apart.

The buyers are more varied than the phrase “police camera” suggests. Flock sells to law enforcement agencies, residential communities and homeowners associations, businesses in retail, healthcare, hospitality, logistics and banking, schools and university campuses, and government entities including parks, transportation departments, and corrections agencies ³. The company’s own product page counts more than 5,000 law enforcement agencies, more than 6,000 homeowners associations, and more than 1,000 businesses among its customers ¹.

That breadth is possible because the hardware asks almost nothing of the site it occupies. Flock’s readers are solar-powered and LTE-connected, and the company advertises that they install in hours ¹ — no trenching, no wiring, no permit for a power drop. A camera can go up at the mouth of a subdivision, on a commercial parking lot, or along a state highway with equal ease.

The cameras are motion-activated and are aimed at the rear of vehicles as they pass ⁴. They do not record continuous video; each trigger produces still images. Municipal documentation is explicit about the limits of what that captures: “Flock ALPR does not record video footage. It is possible that a person may be captured in a still image; however, there is no way to search or filter for a person in the Flock system” ⁴, and the cameras carry no facial recognition and collect no biometric data ⁵.

The images go to Flock’s cloud, where computer vision does the work that makes an ALPR different from an ordinary security camera. It reads the plate characters, identifies the issuing state, and extracts a set of physical attributes the company markets as its patented Vehicle Signature, or “vehicle fingerprint”: make, body type, color, decals, bumper stickers, back racks, top racks, toolboxes, temporary plates, and vehicles with no plate visible ². Direction of travel is logged as well ⁴. Flock’s published performance figures claim more than 99 percent capture in clear and rainy conditions, more than 98 percent at dawn and dusk, more than 96 percent character-recognition accuracy, and more than 97 percent accuracy on the plate’s issuing state ².

Every record is stamped with a time and a location ²— the two fields that turn a pile of photographs into a movement history. ALPRs as a class capture all plate numbers that come into view along with location, date, and time, and can collect thousands of plates per minute ⁸.

The attribute extraction is what makes the archive searchable without a plate number. An investigator working from a partial plate or a witness’s description of a color and body style can search on those attributes alone and compare what was reported against what the cameras captured ². Flock has layered natural-language search on top, marketed as FreeForm, which the company demonstrates with queries like “white sports car with a racing stripe” ¹.

Retention is the last variable. The standard Flock recommendation adopted by most agencies is 30 days, after which records delete automatically unless they pertain to an investigation ^{4 5}; Flock says it follows existing state retention laws where they differ, and that extending retention to as much as a year requires approval by elected officials ¹. Across the wider ALPR industry, the Electronic Frontier Foundation notes that data is often stored for far longer — frequently as long as five years, and by some entities indefinitely ⁸.

One design fact underlies everything else on this page: the camera photographs every vehicle, not only vehicles connected to an investigation. An EFF review of ALPR data found that just 0.05 percent of what the systems collected was relevant to a public safety interest at the moment of capture ⁸.

A single reader is not especially powerful. The network is. Flock cameras check each plate against hot lists in real time: the FBI’s National Crime Information Center database of wanted vehicles ^{1 4}, state and national watchlists, and missing-person alerts including AMBER, Silver, and Missing or Endangered Persons alerts ⁵. Flock joined the National Center for Missing & Exploited Children’s AMBER Alert secondary dissemination system in 2021, and states that within a few minutes of an alert being activated, every law enforcement agency on the network can receive it through the hot list feature ⁶. Agencies can also build their own custom hot lists and share them with other agencies ¹.

Search reaches further than alerting. If a department opts into Flock’s “national lookup” program, its officers can search for a vehicle anywhere in the country where another department or agency has also opted in, and those searches do not require a warrant ⁷. EFF describes the same architecture in general terms: agencies can choose to share their data with thousands of other agencies, and vendors operate the regional and national systems that make cross-country lookups possible ⁸. The ACLU notes that a Flock law enforcement customer can additionally run a search on behalf of another agency that is not itself a customer ⁹.

The practical consequence is that the jurisdiction that pays for a camera is not necessarily the only one that benefits from it — or the one that controls how its record of your street is used. The ACLU cites Syracuse, New York, where police inadvertently shared local plate data through Flock’s nationwide system and it was searched 4.4 million times in a single year ⁹. Illinois offers a measure of the same asymmetry: roughly 90 police departments in the state operate Flock cameras, while more than 461 entities — police departments, prosecutors, dispatch centers, universities, transit agencies, and railroads — had access to Flock data ¹⁰.

Flock describes its own volume as billions of monthly plate reads ¹. The ACLU, working from the company’s figures, puts it more precisely: more than 5,000 American communities, producing more than 20 billion vehicle-tracking plate scans every month ⁹. Growth has been fast and is still uneven across states. In Texas, advocacy groups estimate more than 10,000 cameras statewide, and the number of county and local agencies with data-sharing agreements with the state Department of Public Safety rose from 122 to 206 in five months ⁷.

Neither Flock nor any other vendor publishes a public map of installations, which is why this site exists. More than 127,000 cameras are documented on this site, plotted from records contributed to OpenStreetMap by volunteers; you can browse them by state and city or search your own street on the national map. Two caveats matter. That count includes readers from every manufacturer, not Flock alone, and documentation lags installation — a camera appears only after someone records it, so the absence of a dot is not evidence that a road is unwatched. Our methodology page sets out how the data is collected and where it falls short.

The most consequential document on ALPR accuracy in 2026 came from inside a police department. An audit by the Los Angeles Police Department’s Office of the Inspector General, released in July 2026, examined the stolen-vehicle alerts the department’s Flock cameras generated in August and September 2025. It found 161 vehicles incorrectly flagged as stolen against 337 correct flags — a 32.3 percent false-positive rate ¹². The inspector general noted what such an alert sets in motion: “Often, officers will approach the vehicle with extreme caution or conduct a ‘high-risk’ stop. This involves calling for back up, air support and a supervisor and ordering the suspect out of their vehicle,” adding that “these inaccuracies can affect individual liberty interests, erode public trust, and potentially create substantial legal and financial liability concerns” ¹². The LAPD let its three-year contract expire days later without renewing it, with the department’s chief information officer saying it would discontinue Flock services “until we can get those data, privacy, security and sharing concerns ironed out” ¹².

Errors of that kind arrive by two distinct routes, and the distinction matters for anyone assessing a local program.

The camera reads the plate wrong. Character confusions are the classic failure. In San Francisco in 2009, a reader misread a 3 as a 7 on Denise Green’s plate; officers stopped her at gunpoint and ordered her to kneel and be handcuffed before discovering the error about twenty minutes later, and the city settled for $495,000 ¹¹. In Española, New Mexico, a system mistook a 2 for a 7 on Jaclynn Gonzales’s plate; she was detained at gunpoint and her 12-year-old sister was placed in a patrol car ¹¹. Reads can also cross state lines: in Aurora, Colorado, in 2020, Brittney Gilliam’s Colorado SUV was matched against a stolen Montana motorcycle, and she and four children were handcuffed face-down on the pavement, a case that settled for $1.9 million ¹¹. One study cited by EFF found readers misread the issuing state on one in ten plates, before counting character errors at all ¹¹.

The camera reads correctly and the database is wrong. Here the hardware performs as designed and the hot list is stale. On Thanksgiving 2019 in Contra Costa County, California, Brian Hofer’s car was flagged as stolen after it had already been recovered because the record was never updated; he and his brother were wrongfully detained and settled for $49,500 ¹¹. The pattern recurs. On August 6, 2026, on I-94 near Brookfield, Wisconsin, officers surrounded a car with guns drawn and ordered its occupants out because a Milwaukee homicide alert had never been cleared from the system after investigators no longer needed the vehicle; everyone was released at the scene once Milwaukee police confirmed as much. Milwaukee’s department characterized the episode as “not a Flock camera issue, but a personnel issue” ¹³. The driver said it was the second such stop she had experienced that week ¹³.

The common thread is that a hit is a lead, not a verified fact, and the gap between those two things is where the documented harm sits.

The legal question is unsettled. In January 2026, US District Judge Mark Davis ruled that Norfolk, Virginia’s warrantless use of 176 Flock cameras did not violate the Fourth Amendment, while explicitly leaving the door open: ALPR surveillance “could become too intrusive and run afoul of [constitutional privacy standards] at some point. While a definitive answer to that question is elusive, what is readily apparent to this Court is that, at least in Norfolk, Virginia, the answer is: not today.” The plaintiffs, represented by the Institute for Justice, said they would appeal ¹⁴. A second line of objection concerns who reaches local data: an August 2025 audit by the Illinois Secretary of State found Flock had given US Customs and Border Protection access to Illinois plate data through a pilot program despite a 2023 state law barring the use of such data for immigration or abortion enforcement, and found an out-of-state search run for a woman who had recently had an abortion. Flock’s chief executive, Garrett Langley, acknowledged the company “didn’t create distinct permissions and protocols in the Flock system to ensure local compliance for federal agency users,” and the company paused its federal pilots ¹⁰. State law varies widely on retention caps, sharing restrictions, and audit requirements, so the same camera can be governed very differently on either side of a state line. The practical result has been a wave of municipal reversals: the ACLU counted more than 55 localities that ended Flock contracts in the year to June 2026 ⁹, and we track the individual decisions in our cancellation tracker.

Because no vendor publishes installation locations, the only public record is the one volunteers build. The Finding Flock map plots every ALPR reported to OpenStreetMap in the United States, with the manufacturer, mount, and camera direction wherever a contributor recorded them; the camera index breaks the same data down by state and city so you can see how densely your area is covered relative to others.

If you want to know how your own jurisdiction uses what it collects, the documents cited throughout this article suggest where to look: agencies that operate Flock cameras typically publish a transparency portal listing search volumes and sharing partners, and the retention window and sharing settings are contract terms a council can see and change. To identify the hardware itself, see what a Flock camera looks like. Spot an error in anything here, or a camera we have missed? Get in touch.