Eyes on the Road: AI, Privacy, and Automated License Plate Readers

By Chris Hart and Ariel Chen

Background

The use of Automated License Plate Reader (ALPR) systems is becoming increasingly common in cities across the United States. ALPR systems are computer-controlled cameras sold to law enforcement agencies, which deploy ALPRs at both fixed strategic points and in mobile formats. Although ALPR technology has existed for over two decades (Department of Homeland Security & National Urban Security Technology Laboratory, Automated License Readers Market Survey Report, June 2025, at 1.), the recent integration of artificial intelligence (AI) into ALPR systems raises novel questions about how AI-enhanced surveillance interacts with privacy rights and concerns about discriminatory law enforcement.

Fixed ALPR systems are mounted in specific locations, often on existing infrastructure (such as traffic lights or telephone poles). Peter G. Berris, Kristin Finklea & Dave S. Sidhu, Automated License Plate Readers: Background and Legal Issues, Congressional Research Service, July 21, 2025. Mobile ALPR systems can be mounted on police vehicles, commercial vehicles, or even drones. Id.; Beryl Lipton, That Drone in the Sky Could Be Tracking Your Car, Electronic Frontier Foundation, Sept. 22, 2025. Both systems can capture images of license plates on passing vehicles; bumper stickers; the date, time, location, direction, and lane of travel; and similar information. ALPR systems contain both “real-time” alerts and historical data, allowing users to enter and search for driver information in a database and be notified about new hits relating to specific plate numbers that might be associated with stolen vehicles, missing persons, or suspected criminal activity (known as a hot list). Peter G. Berris, Kristen Finklea & Dave S. Sidhu, Automated License Plate Readers: Background and Legal Issues, Congress.Gov, July 21, 2025.

Traditionally, ALPR systems have used Optical Character Recognition (OCR) technology to capture and identify information on license plates. OCR ALPR systems are typically expensive technologies that use dedicated hardware and can recognize characters on license plates with around 80-85% accuracy. See, e.g., How Do License Plate Readers Work: OCR v. AI Technology, Brite, May 12, 2020.

But ALPRs can also incorporate machine learning and AI technologies, increasing accuracy and lowering costs. Id. For example, ALPR technology can be placed on smartphones to record images and transmit them to driver databases. U.S. Department of Homeland Security, Science and Technology, and National Urban Security Technology Laboratory, Automated License Readers Market Survey Report (June 2025), at 1-2. According to the Department of Homeland Security, “ALPR systems now can read much more than license plates. ALPR software can detect dents on cars, search for specific bumper stickers, process specialty tags, and recognize rideshare logos.” Id. at 1.

Privacy Concerns

Law enforcement’s use of AI-enhanced ALPR systems has raised concerns from civil rights organizations, immigrant rights groups, and litigants focused on privacy. As these actors argue, ALPR systems have created a nationwide surveillance system in which massive amounts of information about individuals can be available to both law enforcement and private parties.¹ AI enhances law enforcement’s surveillance capabilities by increasing the speed and power of analysis and information-sharing.

Civil liberties organizations have pointed out that AI-enhanced ALPR technologies can inappropriately flag so-called “suspicious” travel patterns for use by state and federal law enforcement agencies. Jay Stanley, New Report Highlights How CBP and Border Patrol are Becoming a Repressive Internal Intelligence Agency, American Civil Liberties Union, Nov. 24, 2025. Although ALPRs can be used for legitimate purposes and to promote public safety—such as by collecting tolls or assisting in AMBER alerts—ALPRs enhanced by AI and deployed nationwide can also lead to biased and discriminatory law enforcement. They can be used to target specific groups, such as those who might be suspected of being undocumented immigrants, those seeking reproductive healthcare, or political protestors. Dave Maass & Rindala Alajaji, How Cops Are Using Flock Safety’s ALPR Network to Surveil Protesters and Activists, Electronic Frontier Foundation, Nov. 20, 2025. Advocacy groups note, for example, that ICE has searched local ALPR databases hundreds or thousands of times as part of its enforcement activity. See, e.g., Coalition of Civil Rights and Advocacy Organizations Deeply Opposed to use of Flock Cameras for ICE Surveillance, ACLU Colorado, Aug. 11, 2025. These dangers are exacerbated by the amount of time that information can be kept in law enforcement databases—sometimes indefinitely. Ángel Díaz & Rachel Levinson-Waldman, Automatic License Plate Readers: Legal Status and Policy Recommendations for Law Enforcement Use, Brennan Center for Justice, Sept. 10, 2020.

Legal Framework for Privacy, ALPR Systems, and AI Enhancement

To date, multiple lawsuits have challenged the use of ALPR systems as potentially violative of core federal and state constitutional principles protecting an individual’s reasonable expectations of privacy against warrantless government intrusion. Although few suits have focused on the integration of AI technologies into these systems, key principles have emerged that will inform the evolution of this area in the context of AI.

The Fourth Amendment protects “[t]he right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures.” U.S. Const. amend. IV. In Katz v. United States, the Supreme Court established that the Fourth Amendment “protects people, not places.” 389 U.S. 347, 351 (1967). When a person “seeks to preserve something as private,” and his or her expectation of privacy is “one that society is prepared to recognize as reasonable,” government intrusion into that private sphere generally qualifies as a search and requires a warrant supported by probable cause. Smith v. Maryland, 442 U.S. 735, 740 (1979) (quotations, citations, and alterations omitted). See also U.S. Const. amend. IV (describing probable cause requirement for warrants). Therefore, in general, law enforcement must obtain a warrant when a search would violate a person’s reasonable expectation of privacy.

These foundational principles have taken on new significance in the digital age. The rise of the internet and digital technologies has enhanced government surveillance capacities and raised novel legal questions. Federal and state courts have repeatedly recognized a tension between modern investigative capabilities and preservation of the “degree of privacy against government that existed when the Fourth Amendment” and comparable state guarantees were adopted. Commonwealth v. McCarthy, 484 Mass. 493, 498 (2020) (citation omitted).

In United States v. Jones, for example, Justice Alito reasoned in a concurring opinion that using a GPS device to track a vehicle’s movements over an extended period without a warrant intrudes on reasonable expectations of privacy because “society’s expectation” is that police will not “secretly monitor and catalogue every single movement of an individual’s car for a very long period.” 565 U.S. 400, 430 (2012) (Alito, J., concurring).

In Carpenter v. United States, the Supreme Court concluded that historical cell-site location information (CSLI) enables law enforcement to reconstruct a person’s past movements with such breadth and depth that accessing it constitutes a search requiring a warrant supported by probable cause, describing CSLI as an “all-encompassing record of the holder’s whereabouts” and a “deep repository of historical location information” revealing an “intimate window into a person’s life.” 585 U.S. 296, 311 (2018).

These decisions have considered whether new forms of surveillance exceed ordinary expectations of governmental monitoring to expose intimate details of an individual’s life.

Where Warrantless Surveillance Can Be Found Constitutionally Permissible: Commonwealth v. McCarthy

The Massachusetts Supreme Judicial Court (SJC) has already examined the tension between privacy and emerging technology in the context of ALPRs in Commonwealth v. McCarthy, 484 Mass. 493 (2020). There, the SJC held that the warrantless search of data derived from four fixed ALPRs that captured information from the criminal defendant’s vehicle did not violate the defendant’s constitutional rights. Id. at 512.

In McCarthy, the Massachusetts State Police operated four ALPRs on the Sagamore and Bourne Bridges. Id. at 494. While investigating the defendant on suspicion of drug distribution, the Barnstable Police Department conducted a warrantless search for the defendant’s license plate in both “real-time” alerts and historical ALPR data. Id. The defendant argued that this warrantless search violated his constitutionally protected reasonable expectation of privacy, while the District Attorney argued that the defendant did not have an expectation of privacy in images of his license plate, because the images were from a camera located on a public bridge and the information from the images was observable by the public.

The SJC identified four factors distinguishing information collected by ALPRs from what was observable by “an officer parked on the side of the road”: “(1) the policy of retaining the information for, at a minimum, one year; (2) the ability to record the license plate number of nearly every passing vehicle; (3) the continuous, twenty-four hour nature of the surveillance; and (4) the fact that the recorded license plate number is linked to the location of the observation.” Id. at 508. The SJC stated that the time period in the first factor was long enough to trigger constitutional protections. Id. at 506.

The SJC ultimately concluded, however, that four ALPR cameras located on public bridges (away from constitutionally sensitive locations) provided only limited surveillance that did not divulge the whole of the defendant’s public movements. See id. at 505-09, 512. The SJC noted, nevertheless, that the defendant’s constitutionally protected reasonable expectation of privacy could have been invaded by a more extensive use of ALPR systems. Id. at 508-09.

The SJC clarified that the constitutional analysis “should focus, ultimately, on the extent to which a substantial picture of the defendant’s public movements is revealed by surveillance.” Id. at 506. To that end, the location of the ALPR is crucial. “ALPRs near constitutionally sensitive locations – the home, a place of worship, etc. – reveal more of an individual’s life and associations than does an ALPR trained on an interstate highway.” Id. Additionally, the SJC stated that with ALPRs in enough locations, the “hot list” feature (notifying users about new “hits” for specific plate numbers) could invade an individual’s reasonable expectation of privacy and trigger constitutional protections. Id. at 507.

McCarthy suggests that a new challenge to AI-powered ALPR systems might meet a different fate. Where AI-powered systems depend on data being maintained for longer periods of time and provide complex, detailed, and sophisticated information about individual behaviors, McCarthy’s test would suggest that relying on such surveillance without a warrant could be prohibited or highly curtailed.

Where Warrantless Surveillance Can Be Found Constitutionally Impermissible: Leaders of a Beautiful Struggle v. Baltimore Police Department

In Leaders of a Beautiful Struggle v. Balt. Police Dep’t, the Fourth Circuit addressed the Baltimore Police Department’s (BPD) Aerial Investigation Research (AIR) pilot program. See 2 F.4th 330 (4th Cir. 2021). In 2020, BPD retained a contractor called Persistent Surveillance Systems (PSS) to conduct the AIR program, an aerial surveillance pilot program that was to run for six months with three aircraft flying over Baltimore, covering around 90 percent of the city and capturing about 32 square miles per image every second. Each plane would fly for a minimum of 40 hours per week, amounting to approximately 12 hours per day during daylight hours, weather permitting. Id. at 334, 337.

The photos taken by the planes were transmitted to PSS ground stations, where contractors used the data to “track individuals and vehicles from a crime scene and extract information to assist BPD in the investigation” of certain target crimes. Id. at 334. PSS analysts prepared reports and briefings about target crimes, which could include “from both before and after the crime: ‘observations of driving patterns and driving behaviors,’ the ‘tracks’ of vehicles and people present at the scene; the locations those vehicles and people visited; and, eventually, the tracks of the people whom those people met with and the locations they came from and went to.” Id. PSS was also able to integrate BPD systems into its proprietary software, enabling all systems to collaborate and enhance their ability to solve and deter crimes. Data from the AIR program was retained for 45 days. Id.

The plaintiffs sought a preliminary injunction of the AIR program, arguing that this surveillance violated Baltimore residents’ individual liberties, including the reasonable expectation of privacy. BPD argued that individual physical characteristics would be unobservable in the AIR program because it would not provide real-time surveillance, and because images captured would depict individuals as a single pixel.

The Maryland District Court denied the plaintiffs’ motion to enjoin the AIR program, and a divided Fourth Circuit panel affirmed. See Leaders of a Beautiful Struggle v. Balt. Police Dep’t, 456 F. Supp. 3d 699 (D. Md. 2020); Leaders of a Beautiful Struggle v. Balt. Police Dep’t, 979 F.3d 219 (4th Cir. 2020).

On rehearing en banc, the Fourth Circuit held that, despite the fact that the AIR program had certain gaps in coverage (including because planes would not fly at night or during inclement weather), see Leaders of a Beautiful Struggle, 456 F. Supp. 3d at 704, the warrantless operation of the AIR program violated the Fourth Amendment because it “enable[d] police to deduce from the whole of individuals’ movements.” Leaders of a Beautiful Struggle, 2 F.4th at 346. The Fourth Circuit determined that Carpenter applied squarely to the case. Because data for the AIR program was retained for at least 45 days, the Fourth Circuit stated that this constituted “a ‘detailed, encyclopedic,’ record of where everyone came and went within the city during daylight hours over the prior month-and-a-half,” allowing law enforcement to “‘travel back in time’ to observe a target’s movements, forwards and backwards.” Id. at 341. The court likened AIR data to attaching an ankle monitor to every person in Baltimore. Id. Although the AIR program did not allow for perfect, continuous surveillance, the court found that it enabled “photographic, retrospective location tracking in multi-hour blocks, often over consecutive days, with a month and a half of daytimes for analysts to work with,” which was “enough to yield a ‘wealth of detail,’ greater than the sum of the individual trips.” Id. at 342.

The Fourth Circuit also emphasized that the type of information collected, rather than the type of surveillance technology used, is key to whether constitutional protections are triggered. Id. at 343-44. Whereas the district court dismissed the plaintiffs’ CSLI research study showing that “people’s movements are so unique and habitual, it is almost always possible to identify people by observing even just a few points of their location history,” the Fourth Circuit found this study instructive, stating that “[w]hether those points are obtained from a cell phone pinging a cell tower or an airplane photographing a city makes no difference.” Id.

The Fourth Circuit’s analysis considered the raw data produced by the AIR program and what such data could reveal through deductive reasoning and integration of the data with other law enforcement systems. The court concluded that by integrating police information systems with a highly precise map of movements across 45 days, BPD gained insights about the whole of individuals’ movements. Consequently, the court held that accessing AIR data invades individuals’ reasonable expectation of privacy, constituting a search.

Current Court Challenges

Since Carpenter, Jones, McCarthy, and Leaders of a Beautiful Struggle, litigants across the country have continued to bring privacy- and discrimination-based challenges to law enforcement’s use of ALPRs, highlighting the rapid evolution of this technology and its related privacy implications.

For example, the Electronic Frontier Foundation and the American Civil Liberties Union of Northern California filed a lawsuit in November 2025 challenging the use of ALPR systems under both the Fourth Amendment and the California Constitution’s right to privacy. According to the lawsuit, “[a] person who regularly drives through an area subject to ALPR surveillance can have their location information captured multiple times per day. This information can reveal travel patterns and provide an intimate window into a person’s life as they travel from home to work, drop off their children at school, or park at a house of worship, a doctor’s office, or a protest. It could also reveal whether a person crossed state lines to seek health care in California.”

Similarly, the New Civil Liberties Alliance recently filed a brief in Schemel v. City of Marco Island, No. 25-13913 (11th Cir. 2025), challenging the City of Marco Island’s retention of ALPR records for over three years, which they argue violates Carpenter as a warrantless search in violation of the Fourth Amendment.

The Institute for Justice filed suit challenging the City of Norfolk’s use of over 170 ALPR cameras under the Fourth Amendment. According to the plaintiffs’ filings, most driving routes in the City of Norfolk are surveilled by ALPR cameras, leading the plaintiffs to have allegedly had their information captured hundreds of times. The information is gathered in a database easily searchable by law enforcement, according to the briefing. On January 27, 2026, the district court granted summary judgment in favor of the City of Norfolk. See Opinion and Order, Schmidt v. City of Norfolk, No. 2:24-cv-621, ECF No. 191 (E.D. Va. Filed Jan. 27, 2026). The court distinguished the quantity and quality of data collected by the City of Norfolk from that in Carpenter and Leaders of a Beautiful Struggle, concluding that the collected data points were infrequent and often widely spaced because ALPR cameras were fixed in 75 clusters across the city, recording a vehicle’s location only when it passed one of those ALPRs. See id. at 36-38. The court noted that this level of monitoring “does not ‘follow’ or ‘surveil’ [a] car unless it happens past another fixed ALPR camera.” Id. at 41. The court stated that while police techniques could reveal more sensitive information after an individual becomes the subject of an investigation, it is those techniques, not the ALPR system itself, that reveal private aspects of life. Id. at 4-47. While the court did not conclude that the City of Norfolk’s use of ALPR systems constituted a warrantless search, it did acknowledge that “as the number and capabilities of ALPR cameras expand, the constitutional balancing could conceivably tip the other way.” Id. at 3.

These new case filings should help clarify any limits on governments’ ability to use increasingly powerful ALPR technology for surveillance. In the meantime, absent legislative intervention, it is likely that ALPR systems will proliferate and gain more surveillance power as AI technology improves.

Chris Hart is a Litigation Partner at Foley Hoag, where he co-chairs the Privacy and Data Security Practice. He is currently Vice President of the BBA.

Ariel Chen is a Business Associate at Foley Hoag and works in the Privacy and Data Security Practice.

Endnotes