Grounding the Flock: Confronting Police Surveillance of Mobilities

Abstract

This article explores the integration of AI-surveillance functionality into urban infrastructures to question how the monitoring of mobilities could engender vulnerabilities, particularly for minoritized or stigmatized groups in society. Focusing on the automatic license plate recognition (ALPR) system deployed in the United States by the company Flock Safety, I show how this system acts as a communicative network that exposes people to discriminatory police surveillance. I argue that the network properties of Flock's system facilitate the application of far-right policing cultures across jurisdictions, allowing for unlawful targeting of women seeking reproductive care, immigrants, and vulnerable others irrespective of state or city prohibitions against such profiling. The article concludes by reflecting on countersurveillance efforts to expose Flock's surveillance network and challenge its discriminatory policing logics.

This article explores the integration of AI surveillance into urban infrastructures to question how the monitoring of mobilities could engender vulnerabilities, particularly for minoritized or stigmatized groups. I focus specifically on police adoption of Flock Safety’s automatic license plate recognition (ALPR) system in the United States to target individuals through the proxy of their vehicles. When mapped over time, the movement of someone's vehicle can be highly revealing of their behavioral patterns and social networks: where they live, work, shop, recreate, go to school, seek healthcare, access services, and visit family or friends, and when they do these things. AI functionality translates these many data elements into searchable content and raises obscure patterns to the surface for police scrutiny and intervention (Stanley, 2025). In this respect, Flock's ALPR system contributes to a ubiquitous surveillance environment, one premised on the “knowability” and “addressability” of objects in space and time (Araya, 1995; Murakami Wood, 2008). I argue that the network properties of Flock's system facilitate the application of far-right policing cultures across jurisdictions, allowing for unlawful targeting of women seeking reproductive care, immigrants, and vulnerable others irrespective of state or city prohibitions against such profiling.

Controlling Mobilities

Under the ambit of what has been called the “mobility turn” in the social sciences, scholarly attention has concentrated on the importance of circulation to the constitution of society (Hanam et al., 2006). The movement of people, artifacts, data, capital, and ideas shapes economies, social experience, and life chances (Bennani-Taylor, 2025; Cresswell, 2012). This includes how people make collective sense of their environments and their places within them (Johung, 2016). Because shifts in informational and communication practices structure mobilities, it is vital to trace the ways these shifts are “reconfiguring patterns of movement, co-presence, social exclusion and security across many urban contexts” (Sheller & Urry, 2006, p. 2).

Of particular relevance for this paper's inquiry, policing and surveillance technologies regulate mobilities in ways that are obscured but often consequential. For instance, as will be discussed in further detail, Flock's ALPR system may be deployed to target undocumented immigrants by Immigration and Customs Enforcement (ICE) agents. Used in this capacity, the system recodes mobility as vulnerability; it enacts particular “regimes of mobility” (Glick Schiller & Salazar, 2013) that differentially impact migrant populations, exposing them to potentially violent and dehumanizing forced immobility (detention) or forced mobility (deportation) (Bennani-Taylor, 2025; Pedersen, 2025). In this sense, accelerated movement of data across Flock's network overlays policing geographies upon urban spaces. Data circulation from one jurisdiction to another reconstitutes space (Kitchin & Dodge, 2011). This does not imply that space is remade in the image of abstract computational representation, but rather that bodies within those zones become material interfaces contending with policing data logics. They become aspects of what Azadeh Akbari (2025) has called code/body: an always material body that is subject to wounding from physical and emotional assaults and from forms of undesired data inclusion and exclusion.

Because of automobility's hegemony in the United States, which is supported both infrastructurally and culturally (Seiler, 2009), AI-based surveillance tools to track vehicular movement find a ready context for uptake. Instead of developing in a deterministic fashion, surveillance companies like Flock create a market for their products through a process that Michel Callon (2021) has termed singularization: the adaptation of products to a particular milieu and its users. In Flock's case, this is done by promising police departments and others affordable ways to collect and process millions of data points (e.g., vehicles on roads). AI processing of these data points is built on assumptions about automobility: about the uniqueness of vehicles moving through space, on one hand, and tight correspondences between vehicles and individuals, on the other. In this respect, existing mobility behaviors in the United States influence the AI surveillance being deployed, whereas the situation would likely be much different in countries less invested in automobility or privatized surveillance platforms. As people come to recognize their exposure to such surveillance, they may also alter their personal mobility practices, particularly if they feel themselves susceptible to state interdiction.

Therefore, this article contributes to the mobilities literature by investigating how AI-based surveillance tools foster discriminatory mobility regimes. It posits that Flock's ALPR system affords the circulation of far-right policing cultures, thereby enrolling system users into a regressive policing apparatus designed to criminalize and restrict the mobility of vulnerable populations.

Visibility and Vulnerability

While there are several ALPR vendors operating in the United States, Flock is the largest, with a network of over 83,000 cameras across the country (Cox & Koebler, 2025b) and more than 5,000 contracts with law-enforcement agencies (Flock Safety, 2025a). In addition to law-enforcement clients, Flock also supplies its products to retailers, such as Home Depot and Lowes, and to homeowners’ associations, which implement the system in their communities (Brayne et al., 2024; Koebler, 2025c; McElroy et al., 2021). Police departments regularly access these private cameras too, although, ostensibly, permission has to be granted for them to do so (Koebler, 2025c). Flock cameras can be mounted on vehicles for mobile detection, but the vast majority of the company's cameras are stationary: attached to traffic signals, street signs, light or telephone poles, or highway entrances and exits (EFF, n.d.).

Flock's functionality depends on its pervasive deployment, where every additional contract or camera becomes a “force multiplier” that expands its data collection capabilities and tilts the company further toward monopoly status. Police departments have been drawn to the system because of its affordable subscription-based pricing model, where, depending on the extent of their services, some departments could pay as little as $2,500 per year (McLemore, 2025), although many clients pay hundreds of thousands of dollars for Flock access (e.g., Wray, 2025). As with the normalization of extractive subscription-service platforms elsewhere (Shapiro et al., 2024), surveillance services can be packaged for police to foster long-term dependencies and ensure continued access to data produced by police clients (Gates, 2019). Flock also capitalizes on police-generated data from ALPR-camera deployments to train its AI models and expand its services (McLemore, 2025).

On the surface, Flock may seem like a passive surveillance tool that simply records license plate numbers from proximate vehicles to assist police with solving crimes like vehicle thefts. In actuality, Flock captures a wealth of other data elements, such as a vehicle's color, make, model, and distinguishing features (bumper stickers, dents, roof racks), allowing for the creation of a unique “fingerprint” for every vehicle even if license plates are obscured (DeFlock, 2025; Flock Safety, n.d.). As a Flock promotional document boasts, its Vehicle Fingerprint^TM software supports free-form searches for optimized location tracking: “Instead of manually sifting through hundreds — if not thousands — of images, officers can now search for vehicles with unique characteristics, such as ‘blue SUV with a racing stripe’ or ‘white F-150 with a ladder in the back’” (Flock Safety, 2025c). Flock uses AI not only to respond to police queries but to generate suspicion and send alerts to police when it detects atypical driving habits, such as separate vehicles seemingly driving together or vehicles regularly crossing state lines, which they suggest could indicate drug trafficking (Stanley, 2025).

Flock functions as a communicative network that produces asymmetrical visibility and vulnerability. For instance, records reveal that the Johnson County Sheriff's Office in Texas used Flock in 2025 to search for the location of a woman purported to have had a self-induced abortion (Cox & Koebler, 2025b). Even though abortion before fetal viability is legal in states like Washington and Illinois, the Flock cameras in these states were nonetheless enrolled in the hunt for this woman, which was a search that encompassed 6,809 Flock networks and 83,345 cameras (Cox & Koebler, 2025b). In this example, the affordances of Flock's system seem to encourage such infractions, operating as an accessible search engine—a form of Google for the police (Monahan & Regan, 2012)—that invites abuse while shielding police, for the most part, from scrutiny.

In response to negative publicity generated from this incident, Flock issued several public-relations statements downplaying the abuse (Flock Safety, 2025d, 2026). One related: “The Sheriff stated that [the woman] was not being investigated as a suspect in an abortion-related offense, but was instead the subject of a welfare and safety search. If that account is accurate, the presence of the word ‘abortion’ in the search note does not necessarily mean law enforcement was investigating reproductive healthcare as a crime” (Flock Safety, 2026). However, records obtained by the Electronic Frontier Foundation (EFF) found that the Sheriff's department had opened a “death investigation” for the termination of a “non-viable fetus” and had also consulted with prosecutors about charging the woman with homicide (Maass & Alajaji, 2025). Notably, even after this additional information came to light in October 2025, at the time of this writing in April 2026, Flock had not recanted its earlier defense.

When it comes to immigration enforcement during Trump's second presidency, Flock has also been deployed by federal agents and local police departments to target people for detention and deportation. In particular, local police departments across the country have willingly conducted Flock searches on behalf of ICE agents who did not otherwise have access to the system (Koebler & Cox, 2025). In California, police have conducted searches on those involved in ICE protests and illegally shared those findings as well (Cox & Koebler, 2025a). In other cases, local police have simply given federal agents their login credentials to conduct searches on immigration violations without needing an intermediary (Koebler, 2025b). For instance, this occurred in Illinois, which is known as a “sanctuary state,” where such cooperation with ICE is illegal without a judicial warrant (FAIR, n.d.; Koebler, 2025b). Apart from these “side-door” forms of access to Flock, federal agents have also taken advantage of a Flock pilot program to gain full access to the nationwide camera network and illegally access cameras in Colorado, which is also designated as a sanctuary state (FAIR, n.d.; Koebler, 2025a). Given that ICE raids have occurred at hardware stores like Home Depot (Harter et al., 2025), where undocumented day laborers often look for work, these stores’ inclusion in Flock's network further shows a synergy of public and private data-sharing in support of such immigrant-policing actions.

Out of concern over mounting public backlash, Flock has embarked on a media campaign hinging on denials of the mass surveillance capacities of its network. For instance, one of the company's statements claims: “Flock ALPRs do not and cannot track vehicles, much less individual people. ALPRs take a point-in-time image of the rear of vehicles on public roadways. They are incapable of tracking the whole of anyone's movements” (Flock Safety, 2026). That said, the primary functionality of the system is to facilitate police searches to map the locations of vehicles and their times at those locations, as well as to set “alerts” for when vehicles are spotted by the network. As for tracking people, Flock's latest “Condor” cameras are explicitly marketed to “track people and vehicles, zooming in for identifying details, and capture high-resolution video for investigations” (Flock Safety, 2024). The company's patents make further claims about identifying “different classes of people (male, female, race, etc.) … clothing types (jacket, pants, shorts, hat, etc.) and … height and weight” (Flock Group Inc., 2022, p. 19). Therefore, it is difficult to take the company's defenses seriously when they contradict themselves.

These examples reveal some of the politics of Flock's ALPR system, which cannot be divorced from the punitive, racialized, and often illegal policing contexts of their use. Flock serves as an AI-enhanced communicative assemblage designed to track and regulate others’ mobilities. Used in a routine fashion, the Flock system supports a larger carceral apparatus of racialized policing, one that prejudicially targets people and communities of color. However, situated in a moment of radical disjuncture between states, where reproductive care and immigrants’ rights might be safeguarded in some jurisdictions but not in others, Flock allows for a technological circumvention of norms, values, and laws that would otherwise protect vulnerable individuals.

Resisting Flock

Awareness is growing about Flock's surveillance network, in large part because of creative countersurveillance projects and investigative reporting. One such creative intervention, called DeFlock, is an open-source project dedicated to mapping ALPR cameras across the United States and other countries. The project's website visualizes the prevalence of ALPR cameras in each state by noting their precise location, number, orientation, and vendor, if known.¹ DeFlock builds its camera database through crowdsourcing efforts, and it provides technical resources for visitors to correctly identify ALPR cameras in their neighborhoods and file public-records requests to confirm the systems adopted by their cities.

A few other visualization tools complement DeFlock’s efforts to convey the magnitude of Flock's network. One is Eyes on Flock, a website that compiles data from Flock's rather anemic “transparency portal” to convey total numbers of cameras reported, vehicles tracked, police searches, hotlist alerts, and more.² Through this site, one can ascertain that close to 248 million vehicles have been tracked so far and over 29,000 cameras reported. The site further lists the top reasons given for police searches, the primary organizations with whom data has been shared, and tabulated data for specific cities if the data are known. In a more artistic register, the site flock.ajith.fyi similarly analyzes known audit data from Flock transparency portals to graphically trace—in neon-yellow and blood-red lines—the networks of camera access cutting across the country.³ This site lets users zoom in to determine the number of cameras operated by specific departments, which departments are requesting data from others, and how often. This project simultaneously conveys the breadth of Flock's network and its materialization in specific locales (Kitamura, 2025).

Finally, investigative reporting has been key to unearthing not just the presence of Flock ALPR cameras but also the uses to which they have been put. For instance, one investigator was able to use public-records requests to obtain the Flock audit logs for police departments and discover that the police had given “immigration” or “ICE” as the reason for their Flock searches (Koebler & Cox, 2025). This data trail showed that these departments were assisting with federal immigration policing efforts in violation of Flock's policy and, in some cases, in violation of the sanctuary laws of their jurisdictions. The news outlet 404 Media has been at the forefront of researching and publicizing these disclosures, but civil society groups like the American Civil Liberties Union and EFF have also played an active role through their use of Freedom of Information requests and lawsuits.

In combination, these resistance efforts erode Flock's asymmetries of visibility. They communicate the prevalence of police tracking of individuals’ mobilities and highlight the discriminatory purposes to which that tracking is put. Disclosures have had documentable impact as well, leading, for instance, to cities such as Austin and Denver electing not to renew their Flock contracts and cities such as Lynnwood, Washington, “pausing” their use out of concern about the system being used for unlawful federal immigration operations (Cox & Koebler, 2025a; Jojola, 2025; Lynnwood Times, 2025). Other cities, such as Kalamazoo, have since placed restrictions on the Flock data they share with federal agencies like ICE or Customs and Border Protection (Samples, 2025). Even Flock has audited and revoked some access from those agencies that had illegally facilitated ICE searches (Flock Safety, 2025b).⁴

Conclusion

This article has analyzed Flock's ALPR system as a form of AI-enhanced control over people's mobilities. By operating at the infrastructural level, being layered over existing urban artifacts like streetlights, Flock cameras achieve degrees of relative invisibility, obscuring both their presence and purpose. Nonetheless, the system acts as a potent communicative network that exposes people to dragnet forms of police surveillance. When filtered through Flock's algorithms, mobilities (the act of driving) can exacerbate vulnerabilities, leading to the discriminatory targeting of women seeking reproductive care, immigrants, and people of color, among others. The system betrays affordances for lifting off from local jurisdictional constraints—such as laws protecting immigrants or access to reproductive care—and this allows for far-right policing practices to take place across the country. Moreover, Flock's opacity provides cover for local police to cooperate with regressive or unlawful profiling performed by other law-enforcement entities.⁵ In these ways, the Flock system retrofits existing carceral geographies that are enforced by patriarchal, xenophobic, and racialized policing. In this context, countersurveillance efforts to expose Flock's surveillance network challenge those very policing logics. Resistance visualizes these surveillance networks and more importantly grounds them in specific sites and legal jurisdictions. Such grounding is an important and necessary step in restricting the dangerous gaze of police surveillance.