A Lawsuit Says Flock Built a National Surveillance Network Without Adequate Safeguards

The Facts

On August 25, 2026, Grace Schulte, a resident of Dahlonega, Georgia, filed a proposed class action in the U.S. District Court for the Northern District of Georgia. She sued Flock Group Inc., an Atlanta-based technology company that supplies cameras, databases and search software to law-enforcement agencies.

The cameras are automatic license plate readers, commonly shortened to ALPRs. Unlike a camera triggered by a speeding violation, an ALPR records a vehicle because it passes within view. According to the complaint, each detection can include a plate number, a vehicle image, the time and the camera’s location. Images may also capture drivers and passengers.

Those observations are uploaded to Flock’s databases. An authorized user can search for a plate and retrieve places and times where the vehicle was detected. If agencies share access, the search can reach records collected by cameras outside the user’s own jurisdiction.

A single detection says little beyond where a vehicle appeared at a particular moment. Accumulation changes the nature of the record. Repeated observations may expose routines, associations and visits to sensitive locations, including physical or mental healthcare providers.

The complaint says Flock markets statewide and national search functions as product features. It also promotes tools that identify vehicles appearing at multiple selected locations or traveling near a targeted vehicle. That networked reach is what gives the system investigative value—and what creates the privacy and misuse risks at the center of the lawsuit.

The System Records Its Users, Too

Flock does not simply sell roadside hardware and leave. According to the complaint, it supplies the software and data infrastructure through which users authenticate themselves, search records, share data and generate logs of their activity. Its customer terms also reserve authority to restrict or suspend access in specified circumstances.

The filing says Flock’s audit records can include the username, date, time, stated purpose and plate or other information entered for a search. In practical terms, the platform may contain evidence showing who searched for a vehicle and when.

The person being searched ordinarily cannot see that activity. A driver may have no way to know that an officer queried the plate, even as the system records the officer’s search.

A surveillance platform can preserve proof of abuse after it happens without stopping the abuse while it is happening.

That distinction is fundamental to the complaint. Schulte does not allege merely that individual users broke rules. She alleges that Flock’s prevention, monitoring and intervention controls were inadequate for a system whose misuse could remain invisible to its targets.

The Misuse Cases Cited in the Complaint

The filing compiles reports involving officers, dispatch personnel and former agency employees accused of using license plate systems for purposes unrelated to law enforcement. Arrests and charges are not convictions, and the complaint does not establish the outcome of every cited case.

The numbers become more revealing when searches are repeated. Court records described in the complaint reportedly showed a Florida officer searching for his former girlfriend’s plate at least 69 times, her mother’s plate at least 24 times and her father’s plate at least 15 times.

Another cited case involved a former Braselton, Georgia, police chief who allegedly searched plates associated with a former girlfriend and her teenage daughter roughly 600 times. The filing says a former Richmond County deputy searched one woman’s plate 1,639 times over several months. Three other employees in that investigation allegedly conducted more than 100 unauthorized searches each.

That is the lawsuit’s prevention argument in its clearest form: a log may help reconstruct what happened, but a system that flags suspicious activity only after hundreds of searches can leave the target exposed throughout the abuse.

Access Allegedly Extended Beyond What Customers Intended

The complaint does not confine the alleged problem to rogue officers. It cites examples in which the effective reach of the platform allegedly differed from what government customers believed they had authorized.

In January 2026, according to a City of Mountain View statement described in the filing, federal agencies had accessed data from the California city’s first Flock camera through a nationwide-search setting that Flock allegedly enabled without the police department’s knowledge or permission. A city audit also reportedly found that statewide access did not follow Mountain View’s intended controls.

The complaint says Mountain View sought an explanation, but Flock could not provide historical records sufficient to explain what had occurred.

It also cites an August 2025 announcement from the Illinois Secretary of State. According to that announcement, an audit found that Flock allowed U.S. Customs and Border Protection to access Illinois plate-camera data in violation of state law and lacked proper data-sharing safeguards. That is an agency position reported in the complaint, not a ruling in Schulte’s lawsuit.

A separate incident involved access by Flock’s own personnel. Public-record reporting summarized in the complaint said company employees accessed privately owned cameras connected to Dunwoody, Georgia’s Flock-powered crime system during a demonstration and testing meeting. The feeds reportedly included a children’s gymnastics room, a playground, a school, a Jewish community center and a pool. Flock later said demonstrations would use more public locations, according to the filing.

Together, these episodes support a broader allegation: permissions are not a minor administrative detail. In a networked system, one setting can determine whether a local camera is searchable far beyond the local agency.

At National Scale, Small Error Rates Become Large Numbers

The complaint presents two related but distinct accuracy arguments. First, it attributes to Flock a claim that its cameras correctly capture approximately 93 out of every 100 plates. Applied to 20 billion monthly readings, the plaintiffs calculate that the remaining 7% would equal approximately 1.4 billion inaccurate readings per month.

Elsewhere, the filing discusses a greater-than-96% figure for recognizing individual plate characters under optimal conditions. It calculates that a 4% error rate across 20 billion readings would amount to 800 million inaccurate readings per month.

The filing also cites a real-world review by Roseville, California. The city reportedly examined 1,427 Flock alerts involving vehicles identified as stolen or associated with felonies during 2023 and 2024. It found that 1,011 (more than 70%) contained a misread plate.

The complaint says mistaken plate information has resulted in innocent motorists being stopped, detained, handcuffed, jailed or confronted by armed officers. It relies on reporting about Flock and other automated plate-reader systems for that allegation; the pleading does not provide a judicial finding that every incident was caused by Flock.

Accurate Data Can Still Produce a False Suspicion

A misread plate is the obvious failure mode. The complaint’s more consequential turn is that the plate can be read correctly and still lead investigators in the wrong direction.

A location record establishes, at most, that a vehicle was observed at a time and place. Standing alone, it does not establish who was driving, why the vehicle was there or whether an occupant committed a crime.

The filing points to the experience of Chrisanna Elser in Columbine Valley, Colorado. Flock cameras apparently recorded her truck accurately in an area where a package had been stolen. Because she regularly drove through the area for work, her vehicle repeatedly appeared in the location records. Police accused her of the theft, according to the complaint, and she assembled GPS, vehicle and video evidence to show that she had not committed it.

The alleged error was not faulty optical recognition. It was the inference drawn from ordinary travel.

That matters because Flock markets features designed to identify patterns and associations. The complaint describes Multi-Geo Search, which can identify vehicles appearing at locations chosen by an investigator, and Convoy Search, which can identify vehicles traveling near a targeted vehicle. Other tools permit searches using vehicle characteristics or natural-language descriptions.

Those capabilities may generate leads. They can also turn routine overlap into suspicion: neighbors taking the same road, coworkers arriving at the same workplace, patients visiting the same clinic or strangers moving through the same traffic.

The network does not need to misread a plate to mislead an investigation. Sometimes the mistake begins when an accurate sighting is asked to prove more than it can.

What Flock’s Responses Show—and What They Do Not

The supplied record contains no answer from Flock to Schulte’s lawsuit. It does, however, describe earlier company statements and product changes.

According to the complaint, Flock changed a default retention period for search data from 30 days to seven. The filing argues that a shorter retention period does not stop a user from repeatedly making improper searches while data remains available.

The complaint also says Flock introduced automated monitoring intended to surface atypical search patterns, along with search-term exclusions, case-number functions and sharing controls. Schulte uses those tools to argue that stronger safeguards are technically possible. Whether Flock’s controls were legally inadequate during the proposed class period remains undecided.

The filing quotes Flock Chief Executive Officer Garrett Langley calling abuse “exceptionally rare.” It also reports that, when balancing misuse against investigations the technology allegedly supported, he described the situation as “a trade-off that I can make.” Those statements are presented through an interview cited by the complaint; the full interview was not part of the supplied material.

On accuracy, the complaint says Flock continued to emphasize a greater-than-96% character-recognition rate under optimal conditions after Roseville reported its results. It says Flock blamed the city and claimed performance had improved, while Roseville disputed that characterization.

The evidence in the supplied filing therefore shows a dispute, not a resolved conclusion. Schulte characterizes the company’s changes as selective and insufficient. The court has not determined whether that characterization is correct.

What Schulte Alleges Happened to Her

Schulte lives in Dahlonega, where the complaint says at least 19 Flock cameras operated in September 2025 across two law-enforcement entities. The city’s estimated year-round population was approximately 7,000, according to a municipal source cited in the filing.

The complaint alleges that Schulte’s plate information was repeatedly captured and that she must pass Flock cameras to travel to work. It says changing routes would not allow her to avoid the surveillance.

There is, however, a significant gap in the supplied pleading. It does not identify a particular officer or user who improperly searched Schulte’s plate, provide a date for such a search or describe an audit entry documenting one. The complaint broadly alleges that Schulte and proposed class members suffered unauthorized access or use, but its plaintiff-specific factual allegations focus on collection and exposure to the system.

That distinction may matter. Being recorded by the network, having one’s record searched for a legitimate purpose and being targeted through an unauthorized search are different events. The complaint seeks to connect them through its allegations about the system’s design and the foreseeable risk of misuse. The court has not decided whether those allegations are sufficient or can be proven.

The Five Legal Claims

Schulte proposes a class covering everyone in the United States whose plate data Flock collected through its automated system since January 1, 2024. That is the class she wants the court to approve—not an already certified class.

Class certification is the process by which a judge decides whether one plaintiff may pursue claims for a larger group. The complaint argues that common questions about Flock’s architecture, controls, knowledge and safeguards make collective treatment appropriate. Flock has not had an opportunity in the supplied record to contest that request.

The complaint seeks compensatory, consequential and punitive damages, litigation expenses, attorneys’ fees, interest and other relief. It also demands a jury trial. A request for damages is not an award, and demanding a jury does not mean the case will necessarily reach trial.

What the Court Has—and Has Not—Decided

The supplied record contains a complaint, not a judgment. It does not include an order certifying the proposed class, a ruling that Schulte has adequately stated her claims, a factual finding that Flock’s safeguards were deficient or a determination that the company caused any plaintiff’s injuries.

It also does not include Flock’s formal response. The company could dispute the alleged facts, challenge whether Schulte suffered a legally recognized injury, contest whether the claims can proceed as a nationwide class or argue that agencies and individual users—not the platform provider—controlled the relevant conduct. Those are possible categories of dispute, not positions attributed here to Flock.

The complaint’s references to arrests, accusations and investigations likewise require care. An arrest means authorities alleged that an offense occurred; it is not proof of guilt. News reports can document what officials, records or witnesses said, but they are not judicial findings in this case.

What a Legitimate Fix Looks Like

The complaint identifies three separate control problems: preventing clearly improper access, detecting abnormal use quickly and responding before repeated searches accumulate. Any credible reform would have to address all three rather than treating an audit trail as a complete safeguard.

These are editorial recommendations grounded in the alleged failures described by the complaint. They are not requirements imposed by the court, and the supplied source does not establish which measures are technically or legally mandatory.

The Unresolved Question Is Who Owns the Risk

Individual officers can misuse legitimate access. Agencies can fail to remove former employees or configure sharing correctly. Software can misread a plate. Investigators can draw an unsupported conclusion from an accurate record.

Schulte’s lawsuit attempts to place those failures within one integrated system and assign responsibility to the company that designed and operates important parts of it. Flock’s logs, search tools and access controls are central to that theory because they may show both the power to detect misuse and the limits of what the platform actually prevented.

What remains unresolved is concrete: whether Schulte can prove an injury tied to Flock’s conduct; whether the cited incidents establish a common design failure rather than separate acts by users and agencies; whether a nationwide class can be certified; and whether the law treats large-scale aggregation of ordinary vehicle movements as an actionable privacy intrusion.

The court has answered none of those questions in the supplied record. What the complaint documents is the collision between an enormous searchable surveillance network and safeguards that, according to the plaintiff, too often explain abuse only after the person being watched has already paid the price.