How to Find and Avoid Flock ALPR Cameras Near You

What Flock and ALPRs Actually Are

An automated license plate reader (ALPR) is a camera that photographs every passing plate, converts it to text, and records the plate, location, and timestamp in a searchable database. Flock Safety's own product page describes its cameras as solar-powered, LTE-connected units that install in hours with no wiring, capturing a vehicle's make, model, color, a "Vehicle Signature," the plate, location, and time. Flock states the cameras do not use facial recognition or collect information tied to individuals.

The reach comes from the network. Local agencies build custom "hot lists," and Flock also lets them compare plates against FBI NCIC categories, including an "Immigration Violator" category populated exclusively by ICE. A match triggers an automatic alert, and because agencies share hot lists, a scan logged by one town's camera can surface to any other agency on that list. The Electronic Frontier Foundation documented this cross-agency alerting in June 2026. The EFF has also reported that more than 75% of Flock's police customers opt into a shared national lookup, and that queries against it generally need no warrant.

Scale figures vary by source and date, and no single authoritative live count exists. EFF estimated 80,000 to 100,000 Flock cameras deployed as of early 2026, and some outlets reported the network passing 100,000 cameras by mid-2026. What the network mostly captures is innocent people: the ACLU notes that roughly one in ten reads misidentifies the plate's issuing state, and EFF's review of agency data found only about 0.2% to 0.5% of scans ever match a hot list. The rest are ordinary drivers going about their day.

How to Spot a Flock Camera

Once you know the shape, they are hard to unsee, though no single description fits every unit. According to a teardown by CEHRP, one common Flock unit is a small black rectangular box, roughly nine inches long, mounted eight to twelve feet up on an existing utility or light pole (or a dedicated Flock pole), paired with a small black solar panel and no visible wiring, since both power and data are wireless. They are commonly placed at intersections and highway on- and off-ramps, angled at a traffic lane, and use infrared LEDs with motion-triggered capture rather than continuous recording.

Published descriptions of the housing do not agree

Anyone comparing guides will notice the descriptions diverge. Independent transparency and enthusiast projects variously describe the hardware as a black rectangular box, a dark grey unit about the size of a coffee mug, and a flattened black cylinder, and the mounting heights they give span roughly eight to fourteen feet. No official Flock specification sheet listing exterior dimensions was publicly reachable when this section was compiled, which is a large part of why the street-level descriptions vary. Treat any single precise figure for color, shape, size, or height as one observer's account rather than a spec.

What those independent sources do agree on is a combination of signals, which is the practical way to read a pole:

• A small, self-contained unit paired with its own solar panel and battery.
• Mounted low relative to the street lights around it, low enough to read a plate rather than survey an area.
• A housing aimed along a travel lane, typically at the rear of passing vehicles, rather than angled down across a whole junction.
• No conduit or hardwiring running down the pole, because power and data are both wireless.
• Sometimes a visible infrared illuminator for night capture.

Every one of those signals appears on other equipment too, so it is the cluster that matters, never one item. A guide published by the Finding Flock transparency project puts the ceiling on this plainly: "identification from hardware alone has limits: several vendors sell similar-looking compact solar units, and a pole-mounted camera may turn out to be a traffic sensor or an ordinary security camera rather than a plate reader." A write-up from a company that sells ALPR warning signage makes the same point about the most tempting single tell, noting that "solar panels power traffic counters, weather stations, gate controllers" as well. A solar panel on a roadside pole is a prompt to look closer, not a confirmation.

There is no single Flock camera

That description fits the most common unit, but "what does a Flock camera look like" has more than one answer, because the vendor sells the same capability in several physical forms. Flock Safety's own product page lists a standard reader for two-lane roads and driveways, a long-range reader for multiple lanes and higher speeds, a lightweight relocatable unit it calls Flex, a towable LPR trailer for temporary coverage at events or hot spots, and a video integration product that, in the company's words, turns existing video cameras into readers.

Two of those change what you are looking for. A trailer parked at a roadside for a fortnight is a plate reader even though nothing about it resembles a pole camera. And where the video integration product is in use, a plate read can come from a camera installed years earlier by somebody else for an unrelated purpose, in which case there is no Flock hardware on the street at all. Not finding a black box does not mean the road is not being read.

The long-range unit is worth singling out for what it claims to reach. On its current LPR product page, Flock markets that model as capturing "high-speed vehicles across multiple lanes and long distances, even at 100mph." A reader you pass at motorway speed, several lanes over, is still a reader.

A dome on the pole is a different product

Flock also fields a 360-degree dome camera called the Condor. DeFlock's identification catalogue records it as a dome-type camera with a full 360-degree field of view, tagged as a general surveillance camera rather than as an ALPR, while the plate readers are catalogued the opposite way: fixed, directional, and needing a recorded direction because they are aimed at a specific lane. A rounded dome housing is therefore doing a different job from the rectangular box that may be mounted beside it, and the two are worth logging separately if you contribute to the map.

What a teardown found inside

None of this is visible from the street, but it is the most detailed public account of the hardware and it explains the capability. The CEHRP dissection, by Mike Katz-Lacabe at the Center for Human Rights and Privacy, opened a Falcon-type unit and found a 5MP, 16mm fixed lens on an M12 mount with an infrared filter, a PHD Energy lithium-ion battery rated 10.8V and 19Ah (205.2Wh), a NimbeLink Skywire LTE cellular modem, a Taoglas GPS module, and a WiFi and Bluetooth module, all inside a weatherproof enclosure with connectors at the rear for the antenna and the solar panel. That is an independent teardown by a privacy research organization, not a manufacturer datasheet, and it does not publish exterior housing dimensions. The fixed lens and the single aimed field of view are the useful takeaway for identification: these units are pointed at one lane and stay pointed at it.

Telling it apart from the other roadside cameras

Most poles carry something, so identification rests on the combination of features rather than any single one. The plate-reader signature is a small solar panel with no conduit running down the pole, a housing aimed along a lane rather than down at a junction box or across a whole intersection, and a mounting height low enough to read a plate rather than high enough to survey an area. As a practical heuristic, signal-control and traffic-counting equipment tends to be mains-wired with a cabinet at the pole base, and red-light or speed-enforcement installations tend to be physically larger and mains-powered. Those are rules of thumb for narrowing down a pole, not identifications in themselves.

A few contrasts help in the field, offered in the same spirit of rules of thumb. Pneumatic-tube traffic counters lie across the road as rubber hoses anchored at the roadside, so nothing about them resembles a pole-mounted camera at all. Video and radar traffic counters are generally mounted high and angled down over an intersection to watch the whole junction rather than sighted along one lane, which is the opposite geometry to a plate reader. Red-light and speed-enforcement cameras are usually larger, mains-powered, and tied to a specific signal phase or approach. These distinctions come from independent field observation rather than from a transport-agency specification, so use them to narrow a guess rather than to settle one, and pair them with the lane-aimed, low-mounted, solar-and-no-conduit cluster described above.

One caution worth taking seriously: the absence of a logo proves nothing. Housings are frequently unmarked, and the DeFlock catalogue tracks equipment from more than thirty manufacturers, a good deal of which looks broadly similar at a distance. If you are logging a camera you cannot positively identify, record it as a generic ALPR rather than guessing the make. A wrong make on the map is worse than a missing one.

Map Them With DeFlock

The community project DeFlock (deflock.org, which the older deflock.me redirects to) crowdsources ALPR camera locations onto OpenStreetMap. Volunteers photograph cameras and add tagged map points, and the live map at maps.deflock.org renders directly from OpenStreetMap data. OpenStreetMap held 138,351 features tagged with a surveillance type of ALPR worldwide as of 9 August 2026, spanning Flock and other vendors, which makes it the largest crowdsourced ALPR registry available. The dataset is weighted heavily to the United States, with 127,878 of those in the US extract. A figure of 336,000 circulates for this dataset and does not belong to it: that number is Flock Safety deployment, roughly 75,000 US cameras plus 261,000 internationally as mapatlas.eu itself breaks it down, and its international weighting is the reverse of what the crowdsourced map looks like. The project requires no login to browse, and states it uses no tracking, analytics, or ads. Power users can query the raw OpenStreetMap data directly through the Overpass API (searching for nodes tagged with a surveillance type of ALPR) to pull camera points for a whole area at once.

The map exists because someone refused to be quiet about it. Creator Will Freeman started DeFlock in 2024, and when Flock sent a cease-and-desist over its trademark in late 2024, the EFF represented him pro bono and he declined to take it down. Checking DeFlock for your area is the fastest way to see which routes near you are already covered.

Other Ways to Find Cameras Near You

DeFlock's map is the broadest starting point, but it is volunteer-built and therefore uneven from place to place. Two other routes are worth checking for your own town, and both come with a hard ceiling on what they can show.

Agency transparency portals

Flock offers its law-enforcement customers an opt-in public page, published at addresses following the pattern transparency.flocksafety.com/[jurisdiction-name], and some agencies also link it from the city website. Security engineer Ryan O'Horo maintains an independent aggregated list of 562 known transparency-portal URLs, which is the quickest way to find out whether the agency covering your area publishes one at all. His own caveat is the important part: "Only a small number of law enforcement customers enable this feature (< 10%), which implies that there are thousands of other organizations for which we cannot directly view how they use Flock Safety data." The ACLU describes what the portals that do exist actually publish as barebones information. A portal is a starting point for a records request, not a substitute for one.

Aggregators built on those portals

A project at Eyes On Flock describes itself as "an independent project aggregating and analyzing data from Flock Safety's public transparency portals," tracking cameras, searches, and data sharing across the United States. Because its input is the same opt-in portal system, it inherits the same limitation: it can only see the agencies that chose to publish, which is under one in ten of them.

The ACLU's own toolkit

The ACLU's Get The Flock Out toolkit points residents at both DeFlock and the Have I Been Flocked search covered in the next section, and pairs them with the steps that actually produce records: public-records requests and turning up at council meetings where contracts are discussed. Taken together, these are the closest thing available to a public inventory. No complete list of Flock camera locations exists anywhere in public, and any tool claiming to be one is overstating what the underlying data allows.

Have I Been Flocked, and What That Tool Can Actually Tell You

"Where are the cameras" and "has anyone looked me up" are different questions, and a tool at haveibeenflocked.com answers a narrow version of the second one. It searches Flock audit logs that have been released through public-records requests and reports whether your plate appears in one of those logged searches. The ACLU's toolkit lists it as a research tool alongside DeFlock.

Start with the limit, because the limit is most of the story. It cannot tell you whether a Flock camera ever photographed or scanned your car. In the site's own words, the results "don't show when or if your vehicle passed one of Flock's estimated 120,000+ cameras and connected devices, or what those devices returned." It shows searches, not scans. Nothing public exposes the scan record for an individual vehicle. (That 120,000-plus figure is Flock's own deployment estimate as the tool cites it, counted by a different method from the crowdsourced OpenStreetMap tally quoted earlier, so the two numbers are not directly comparable.)

A hit is also narrower than it first reads. Per the tool's own explanation, "a search result means an operator on the Flock system queried the database for your license plate using the Flock application/website. This user is not necessarily a police officer, and a search does not imply you were stopped or investigated." It adds that search policies vary and that many police agencies require neither a warrant nor suspicion to run one.

A miss tells you less still. The dataset covers only the audit logs that have actually been disclosed, few governments provide easy access to them, and the records that are released are often redacted. There can also be a long lag, running to months or years, between a search happening and it appearing in the dataset, which the site sums up as "this is not a real-time monitoring tool." No result therefore means no matching record in an incomplete and stale public dataset, not that nobody searched for your plate.

On the query itself, the site states that it "does not save, store, or log the license plate numbers or search terms you enter," and that the search runs and the data is immediately discarded. Its live counters, which change continuously, showed roughly 4.68 million plates across about 242 million logged Flock searches when checked on 16 August 2026; treat those as a snapshot of how much has been pulled loose so far rather than as a fixed statistic.

The Data Rules, and Why There's No Real Opt-Out

Here is the hard truth: there is no national opt-out registry for ALPR data, and no consumer-facing way to remove yourself from Flock. Flock positions itself as a data processor, not the data owner. When a California resident invoked their CCPA rights in April 2026 and demanded deletion, Flock replied that the request "cannot be completed" and redirected them to the specific law-enforcement agency that owns the camera data.

Per Flock's stated policy, the standard data retention has been 30 days, after which footage is automatically deleted "unless local law says otherwise," extendable up to a year with an agency's approval. Flock frames data sharing as "a local choice" controlled by the contracting agency. That 30-day default is now changing, though not everywhere at once, and the difference matters if you are trying to work out how long a record of your car survives.

What Flock announced in August 2026

On 13 August 2026, Flock announced it is shortening the standard retention window from 30 days to seven, with anything held longer than that required to be evidence tied to a case number under what the company calls Evidence Mode. The change applies to new configurations: existing customers keep the retention periods already agreed in their contracts, so this is not a network-wide switch on day one, and the camera network your town runs may still be on a 30-day or longer clock. The announcement was reported by the Associated Press via Fortune, and our own coverage breaks down the case codes, the audit lockout, and the shorter retention default.

The same announcement turns three previously optional controls into requirements:

• Mandatory case codes. Officers must enter a code from their records management system tying each search to a specific case before running it, and emergency overrides are automatically flagged for review.
• Mandatory audit assistance with lockout. Every customer must run an audit tool intended to flag abnormal search behavior. When it triggers, the user is locked out pending an internal review. The tool existed before this announcement but was optional, and roughly a third of customers had adopted it.
• City-level offense controls. Agencies can decide which offense types, such as homicide or arson, outside agencies may search their data for. That is the mechanism a city would use to block outside searches related to immigration enforcement.

Two caveats belong with all of this. First, the timing: the mandatory compliance deadline is 1 January 2027, so as of writing these guardrails are announced and rolling out rather than universally in force, and a search run today may still predate any of them. Second, the reception: the ACLU and the Center for Democracy & Technology publicly dismissed the changes as inadequate, as covered in our report on the announcement. A shorter default retention narrows the window in which a record of your car exists; it does not change who can search it or on what basis.

That leaves two real levers:

• Request records or deletion from the agency, not Flock. The police department operating a given camera is the data controller. In states with strong privacy law (California's CCPA, for example), a request to that agency may carry legal weight that a request to Flock does not.
• Route avoidance. Data can only be collected on a plate that passes a camera. Using DeFlock to understand where the cameras sit is the only practical way to reduce how often your movements are logged.

One lever that people reach for but that does not help here is the DMV or data-broker opt-out. Under the federal Driver's Privacy Protection Act of 1994, state DMVs sell bulk driver records to brokers, and per an InvestigateTV review, only Delaware, Wisconsin, and Wyoming offer any consumer opt-out at all. Even those carve out law-enforcement, insurance, and employment uses, and none of it touches ALPR data anyway, since ALPR records are collected by cameras on the road, not pulled from DMV files.

Reduce Exposure Through Legal Choices

The only reliable individual lever is awareness of where the cameras are and choosing around them where you can. Use DeFlock to see which roads near you are covered, then, where it is practical, pick a route, or take transit or a rideshare, that passes fewer cameras. This will not make you invisible, and for daily commuting it may change little. It simply reduces how often and how completely your movements are logged.

It helps to know where ALPRs cluster. The common placements are intersections, highway on- and off-ramps, and the main entrances to a town, all chosen to catch the most traffic. Private cameras add to the picture: homeowner associations, apartment complexes, and parking operators increasingly run their own Flock or similar units, and many of those feed the same shared network as the ones the police install. Treat a gated community or a large parking structure as monitored by default.

Beyond Your Car

Plate cameras are one layer. A few related pieces are worth knowing, both to understand the wider system and to see where public pressure has already changed it.

• The cancelled Ring and Nest deal. In October 2024, Amazon announced a doorbell integration that would have connected Ring and Nest cameras into Flock's police-request pipeline. After sustained backlash, including a Super Bowl-timed campaign and a letter from Senator Ed Markey, Amazon cancelled the deal in February 2026 before it ever went live. It is a clean example of organized resistance working. See our coverage of the cancelled Ring and Flock partnership.
• Detecting smart glasses. A free app called Nearby Glasses (from developer Yves Jeanrenaud) listens for the Bluetooth advertising signals broadcast by smart glasses such as Ray-Ban Meta and alerts you when one is close, as reported by TechCrunch. It is on Google Play, and an iOS build shipped in 2026, though Apple's background limits mean the iOS version only scans while it is open. It cannot see glasses with Bluetooth switched off, so treat it as a partial heads-up rather than a guarantee.
• Fusion centers. Roughly 80 DHS-linked fusion centers buy ALPR feeds alongside other data streams and share them across agencies, per the DHS fact sheet, and the ACLU has documented misuse. This is the plumbing that lets one town's cameras feed a far larger system.
• Dashcams, framed correctly. A dashcam does not detect or block ALPRs, and it will not keep you off any list. It is a documentation tool. If you are pulled over because of a misread plate, which is a common and documented error, your own footage can help establish what actually happened. Think of it as evidence, not defense.

Civic Action That Actually Works

This is the part that moves the needle. Nearly every documented reform started not with a gadget but with a records request and a city-council meeting.

• File public-records requests with your local police. The single most effective step is a public-records or FOIA request to the department that runs the cameras. Ask for the Flock contract, any data-sharing agreements with DHS, ICE, or CBP, and the access logs showing federal searches. Those records are what expose policy violations. In November 2025, a Washington court ruled that data captured by Flock cameras are public records subject to disclosure, a precedent that makes those logs harder to withhold.
• Push for a CCOPS ordinance. Community Control Over Police Surveillance ordinances, built on an ACLU model, require public notice and council approval before police can acquire surveillance tech, including ALPRs. Around 26 jurisdictions have adopted one, from Madison to St. Louis, with the model originating in San Francisco.
• Show up when cameras are up for a vote. Cities have pulled the plug once residents pushed. Sedona, Arizona voted unanimously to reject Flock in September 2024 in an early, high-profile cancellation; Austin, Texas declined to renew in 2025; Oak Park, Illinois voted to end its program in August 2025; Evanston, Illinois switched off 19 cameras the same month; Bend, Oregon moved to cancel in January 2026; and Mountain View and Santa Clara, California scaled back. The city-win playbook is detailed in our sanctuary-cities coverage.
• Know your state's law. At least 16 states now regulate ALPRs in some way. Illinois, in 2023, was the first to bar sharing ALPR data for abortion or immigration enforcement; Washington's SB 6002 pairs driver-privacy rules with its plate-cover ban; Colorado's SB26-070 pushes a warrant requirement; and Kentucky's HB 58 caps retention at 90 days.
• What a detailed state law looks like, in one state. Washington's SB 6002 took effect on 30 March 2026, and a summary by the Municipal Research and Services Center of Washington sets out how far it reaches: restrictions on ALPR collection at healthcare facilities providing protected services, immigration facilities, schools, places of worship, courts, and food banks; tiered retention limits, with a 21-day default, 12 hours for parking enforcement once a matter is disposed of, six months for commercial-vehicle enforcement, and 30 days for traffic studies; and a requirement that state agencies register with the Washington Attorney General by 30 September 2026. This is Washington law and Washington only. None of it applies in another state, and it is included here as an illustration of what a legislature can specify, not as a description of your rights wherever you live.

The 2026 Backlash

The system is under real pressure, though the exact scoreboard depends on how you count. A GovTech analysis counted 82 Flock contracts terminated across 28 states between August 2021 and May 2026, with 39 of those in the first five months of 2026 alone. Other tallies count cities rather than contracts, landing in the range of 50 or more (EFF) to roughly 80, so treat any single figure as one method, not a settled total. A California class action alleges Flock shared ALPR data with out-of-state and federal agencies in violation of state law, citing SFPD data being searched from out of state more than 1.6 million times over seven months, per KTVU.

Accuracy is part of the story too. In July 2026, an audit reported by TechTimes found about one in three of the LAPD's hot-list alerts were false, and the department cut off Flock. The throughline behind most cancellations, though, is federal data access. In Dayton, Ohio, an audit found the city's Flock cameras had been searched more than 7,100 times for immigration-enforcement purposes in violation of city policy; the city suspended the data and physically covered the cameras with black trash bags. And an EFF report documented a Texas deputy searching more than 83,000 cameras to track a woman over a suspected self-managed abortion, after which Flock added keyword filters blocking "abortion" and "immigration" as search reasons.

Two accuracy numbers that are constantly confused

Two more figures circulate in coverage of ALPR accuracy, and they are routinely mashed together into a single claim that neither of them supports. They measure different things, on different denominators, and should be read separately.

• More than 96%, character-level. Flock claims, as reported by Yahoo News, that its cameras can accurately read more than 96% of license plate characters. That is a claim about optical character recognition on individual characters, not about whether the resulting alert is right.
• 71%, whole alerts in one city. Roseville, California police reviewed 1,427 stolen-vehicle and felony alerts from 2023 and 2024 and found 71% of them were misreads, per the Davis Vanguard. As that report is careful to state, the figure "does not mean that Flock incorrectly read 71% of every license plate captured by its cameras." It is the error rate within that specific set of reviewed alerts in that one department.

Combining them into one headline accuracy rate produces a number that exists in no source. It is also worth carrying the department's own qualifier: Roseville police said they were not aware of any wrongful stops or arrests in the city resulting from the inaccurate alerts, because officers are required to independently verify a Flock alert before taking enforcement action. That verification requirement is exactly the sort of policy detail a records request to your own agency can establish, or fail to.

Related Coverage

• The Flock and ALPR master index: every report on this site about plate readers, routed by question.
• Flock's August 2026 guardrails: case codes, audit lockout, and a seven-day retention default
• Sanctuary cities and the Flock cancellation playbook
• How plate data feeds broader vehicle tracking (OSINT)
• How often ALPRs are wrong, and who gets hurt
• Flock cameras found exposed online without a password
• The San Jose warrantless-ALPR lawsuit
• The cancelled Ring and Flock partnership
• The federal car-surveillance mandate
• The full guides hub

If your threat model is higher than the average driver's, our guide to a surveillance-resistant digital life goes further into OPSEC, though most readers will not need that depth.