By the time many residents learn about Flock Safety cameras in their communities, the devices are already tracking every passing vehicle — including theirs. With little warning and minimal public discussion, these AI-powered license plate readers have proliferated across the United States, quietly building a surveillance network that raises profound privacy concerns.

Flock Safety, a company based in Atlanta, markets its cameras as a tool for law enforcement, businesses, schools, and homeowners associations. The company claims more than 12,000 customers use its products, and public estimates suggest roughly 90,000 Flock-branded cameras operate nationwide. The cameras are easy to overlook: small black boxes attached to utility poles, often powered by solar panels or batteries, connected via cellular networks. They resemble ordinary traffic cameras and can be mistaken for infrastructure monitoring devices.

What Flock cameras capture

The most common model, Flock Falcon, does not measure speed. Instead, it takes six to twelve still images every time a vehicle passes. Using automated license plate recognition (ALPR) software and machine learning, the camera extracts a vehicle's license plate number, make, model, color, body type, direction of travel, visible damage, aftermarket wheels, bumper stickers, and other distinguishing features. These records are not stored on the camera itself; they are transmitted over cellular networks to Flock's cloud system, hosted by Amazon Web Services. Authorized users — typically police departments — can search the data through a dashboard and set up alerts for vehicles linked to stolen plates, missing persons, arrest warrants, or other databases.

Flock also offers a pan-tilt-zoom camera called Condor, which can detect and follow humans using a feature called Guardian Mode. The company maintains that none of its cameras use facial recognition, but the Condor's ability to track people raises additional surveillance concerns. Investigative reporters have demonstrated the Condor following a reporter as they walked, highlighting its tracking capabilities.

The risks of a shared network

A single Flock camera captures a snapshot of a vehicle at a specific time and place — similar to what any bystander might see. The danger lies in the network effect. When hundreds or thousands of cameras across towns, states, and regions feed data into a single searchable system, law enforcement can reconstruct a person's entire day, week, or month of travel patterns. Catherine Crump, a clinical professor of technology law at UC Berkeley, explains that aggregated location data can reveal where someone lives, works, worships, and socializes. Unlike traditional surveillance that requires real-time observation, this historical trail can be searched at any time, allowing investigators to trace movements long after they occurred.

AI amplifies these risks. Instead of manually reviewing images, operators can use artificial intelligence to search for patterns untied to specific crimes. For example, police might query the system for vehicles seen near a protest, even if no crime has been reported. Such proactive use raises concerns about profiling, inaccurate conclusions, and mission creep. Trevor Horwitz, co-founder of cybersecurity firm TrustNet, notes that AI can infer behavior that may not be criminal, leading to suspicion without evidence.

Data security is another concern. All records are stored in the cloud, which, while not inherently insecure, becomes a target for hackers. More insidious is the risk of insider abuse. Noam Birnbaum, CEO of cybersecurity firm Ignition, points out that the greatest danger often comes from people already granted access. Once data is shared across agencies — which Flock's system allows — local protections may become meaningless. A city ordinance that restricts data use to specific purposes cannot prevent another agency that downloads the information from using it however it wishes.

False alerts: When AI gets it wrong

The consequences of flawed AI are not hypothetical. In 2025, Joel Feder, director of content at The Drive, was driving a Range Rover with a manufacturer plate through Minnesota when police surrounded him and his wife. Flock's cameras had captured a partial reading of his plate, which resembled one reported stolen in California. The AI incorrectly linked the two, triggering an alert. Feder recalls officers with hands on holsters, ready for confrontation. The incident resolved without harm, but it illustrates the immediate danger of over-reliance on automated systems.

Feder argues that Flock and other ALPR companies must improve guardrails. A partial plate match should not be treated as an exact match. Systems should flag uncertainty, requiring human verification before action. Feder told reporters that technology has advanced far faster than regulations, and new laws are urgently needed to govern use, access, and oversight.

Speaking out against Flock

Former Rhode Island police officer Noel Pichardo publicly criticized Flock cameras after testifying before state lawmakers about their lack of public consent. He claims his career ended because he raised concerns. The Rhode Island Police Chiefs' Association disputed his account, but Pichardo's videos drew widespread attention. He told reporters that he wanted residents to know cameras were installed without their knowledge or consent, and that intelligent debate about their use is needed before they become permanent.

The town of Saranac Lake, New York, provides a counterexample. Residents protested after Flock cameras were installed with little warning. The village council held public hearings, and the contract was eventually canceled. The cameras came down. But in most communities, such discussions never happen. Once cameras are in place, they often stay.

How to find Flock cameras and advocate for transparency

A volunteer-run website, DeFlock.org, maps reported Flock camera locations across the US. Users can search by address, view photos, see camera models, and the direction they face. A route tool estimates how many license plate readers a trip may pass. The map is maintained by volunteers, so some entries may be outdated. For example, Saranac Lake still shows cameras on the map even after removal.

After identifying a nearby camera, residents can contact their local police department, town office, or property manager to determine who owns and operates it. If ownership is established, residents can request copies of contracts, data retention policies, data-sharing agreements, audit logs, false alert procedures, and a list of outside agencies with access. They can raise concerns at public meetings, contact elected officials, file Freedom of Information Act requests, or demand hearings before additional cameras are installed.

Civil rights attorney Adanté Pointer emphasizes that there must be robust public debate to determine Flock policies, especially regarding data sharing and access. Without discussion, residents are left unaware that their movements are being recorded, analyzed, and stored in a system they never consented to.

Flock Safety did not respond to questions about its practices before publication. The company continues to expand, with more than 90,000 cameras now active. Whether residents have a say in their installation depends largely on how proactive local communities are in demanding transparency. The case of Saranac Lake shows that public pressure can succeed, but it requires awareness first. Many towns, like the one described by the reporter who first broke this story, are only learning about the cameras after they are already up and running.

Source:ZDNET News