By John Ellis¹
Editor’s Note: This is the third article in our series on Chatrie v. United States. The first explained the Supreme Court’s holding that obtaining Google Location History through a geofence warrant is a Fourth Amendment search. The second applied that reasoning to cellular area searches. This article turns to roadside camera networks.
“Automated license plate reader” is an increasingly misleading term. Modern cameras do not merely read license plates. They photograph vehicles, record where and when they appear, classify their make, model, color, and distinguishing features, aggregate records across multiple jurisdictions, and feed searchable systems that can reconstruct movements and infer associations. The license plate is only the index to the conglomerated data.
The surveilling network may be much larger than just the cameras owned by one police department. Depending on sharing permissions, officers may be able to search records generated by cameras operated by other agencies and participating private systems well beyond their own jurisdiction. Businesses, homeowners’ associations, and other private customers may also operate cameras whose records are shared through commercial platforms. The relevant system is therefore not simply the roadside camera. It is the camera, cloud database, sharing network, and search tools working together.
One camera recording a visible plate on a public road will not ordinarily present the strongest Fourth Amendment issue. But the network the camera is tied to is doing much more. The strength of the Fourth Amendment issues comes from government’s later query of the aggregated, searchable vehicle-location database those cameras create. When officers reconstruct a vehicle’s history, identify every vehicle near a place, or generate leads from patterns and associations, the investigation begins to resemble the retrospective location searches addressed in Carpenter and Chatrie.
What Happens After the Capture
A camera captures a plate and the vehicle, which may include its occupants. Software adds time, location, direction of travel, and vehicle characteristics. The record is stored, sometimes shared across agencies and vendor networks, and later searched. The result may be a photograph or video. It may also be a travel history, a list of every vehicle near a location, a recurring pattern, an association between vehicles, or a new investigative lead. That range of uses matters because these cameras can operate continuously, including at night, without requiring officers to choose a target before the observation occurs.
Four different investigative uses can hide beneath the label “plate reader.”
- A real-time hot-list alert is the narrowest. A known wanted or stolen plate passes a camera and the system alerts an officer. This looks most like technologically enhanced observation of a known vehicle in public.
- A historical plate search retrieves sightings of a known vehicle over days, weeks, or months. This is the closest Carpenter analogue: the government searches a stored location record to reconstruct past movements.
- An area-and-time search identifies every vehicle that passed a location during a selected period. A multi-location version compares several crime scenes and finds vehicles common to them. This is the closest Chatrie analogue: search the many to find the few.
- A pattern or association search identifies vehicles that repeatedly appear together, visit selected locations, or match a physical description. It may ask, for example, which vehicles repeatedly appeared in close proximity to a suspect’s vehicle across multiple places and times. The system does more than retrieve observations; it generates investigative leads from a population’s movements.
Lumping all four uses together as “plate reading” conceals the constitutional problem. A public observation and a retrospective dragnet are not the same government act.
Tiny Constables, Now Networked
United States v. Jones supplied a memorable image. Discussing the eighteenth-century analogue to GPS tracking, Justice Alito joked that sustained surveillance would have required “a gigantic coach, a very tiny constable, or both.”² The joke captured a practical limit. Long-term surveillance once required people, vehicles, time, and money. Roadside camera networks remove much of that limit. No tiny constable rides in the car. Cameras record passing traffic automatically, software remembers the observations, and an officer can search the database later—without having decided whom to follow when the travel occurred.
That is why United States v. Knotts does not end the analysis. Following one vehicle during one trip is different from querying a preexisting database capable of reconstructing multiple trips or identifying vehicles that were unknown to police when the observations were collected. Carpenter recognized the constitutional importance of that shift from ordinary observation to effortless retrospective tracking. Chatrie applied the same concern to a database searched by place and time.
Vehicle location data is not identical to cell site location information or Google Location History. A roadside camera network does not ordinarily follow a person indoors, and a sparse network may yield only occasional sightings. But as coverage, retention, sharing, and analytics grow, so does the capability the Court has repeatedly found constitutionally significant: cheap, retrospective reconstruction of movements that conventional surveillance could not realistically achieve at the same scale.
Why Chatrie Matters
Three parts of Chatrie are especially important.
First, the stronger Fourth Amendment issue may lie in access to the historical database, not in the individual roadside image. Chatrie held that officers conduct a search when they access a person’s Google Location History. It did not hold that every database query is a search. But it reinforces the argument that the analysis cannot stop with the public nature of each isolated observation. A court must examine the database the government accessed, the scope of the query, and what the search was capable of revealing. The Massachusetts Supreme Judicial Court anticipated that point in Commonwealth v. McCarthy. Four cameras at two bridges produced too limited a record in that case, but the court recognized that, “[w]ith enough cameras in enough locations,” historical plate-reader data would invade a reasonable expectation of privacy.³
Second, a short window is not a categorical exemption. Chatrie rejected the government’s proposed grace period for obtaining protected location data. Duration and density still matter here: one image from a sparse network is not the same as months of travel history. But a court should not resolve the threshold question solely by counting hours or asking whether the system captured the “whole” of a person’s movements. Even a short area search may expose a visit to a home, medical office, attorney, place of worship, or political gathering—and may search the records of many uninvolved people to identify one suspect.
Third, the third-party doctrine is weak, and sometimes irrelevant. If police own the cameras and database, there may be no third party at all. With a commercial network, the government accesses vehicle-location records drivers did not voluntarily provide. A plate is displayed because the law requires it, not because a driver agreed to create a searchable account of travel. If enabling Google Location History did not amount to voluntary exposure in Chatrie, ordinary driving should not be treated as consent to perpetual aggregation.
The Search Identifies Its Own Targets
The deepest problem with area, multi-location, and association searches is architectural. A conventional warrant generally identifies its object before execution: a person, place, account, device, or item. A dragnet database query begins with a geographic or behavioral criterion while the people whose records will be examined remain unknown. Which vehicles passed through this intersection? Which appeared near three burglaries? Which vehicles repeatedly appeared in close proximity to a suspect’s vehicle across multiple locations? The database supplies the identities afterward. The government uses the query itself to decide whom to suspect. That is the same inversion at the heart of a geofence warrant. It places probable cause, particularity, and officer discretion squarely in issue. “Query” is useful terminology here. It describes the technical operation the officer performs on the database. The constitutional question is whether that query constitutes a Fourth Amendment search.
Four Questions for Suppression Litigation
Once a historical or dragnet query implicates the Fourth Amendment, four questions frame the suppression litigation.
- Probable cause. Why was this particular query likely to find evidence? Probable cause that a crime occurred near a road does not necessarily justify searching the records of every vehicle that passed. Probable cause to investigate one vehicle does not automatically justify reconstructing months of travel. The analysis should examine the fit between the known facts and the cameras, geography, time range, networks, filters, and analytical tools used.
- Particularity. The warrant should identify the databases and networks to be searched, the permitted query types and filters, the information to be returned, the criteria for identifying an owner or driver, and the handling of unrelated records. A warrant that lets officers enlarge the area, lengthen the period, alter the description, or reach new networks without returning to a magistrate judge risks becoming a general warrant.
- Execution and reasonableness. What did the system actually search? Which cameras, agencies, and jurisdictions were included? How many vehicles were returned? Did officers modify filters, run follow-up queries, or review data outside the authorized period? Were unrelated results retained, exported, or reused? These are discovery questions, not facts to assume.
- Good faith. Under United States v. Leon, reliance may not be reasonable if the application concealed the network’s scope, omitted shared access or analytical tools, or left the search’s real boundaries to the vendor or executing officers. Davis v. United States separately concerns reliance on binding appellate precedent. Decisions approving a narrow alert-driven use should not automatically authorize historical, geographic, or associative searches of a much larger network.
Practice Pointers
Start with the query, not the camera. Determine whether the evidence came from a real-time alert, a known plate’s history, an area-and-time sweep, a comparison across locations, a physical-description search, or an association tool. The constitutional argument will depend heavily on that distinction.
Then obtain the full record. Discovery should identify every query and parameter; the users and agencies involved; the locations of the cameras reached by the query; the jurisdictions and networks searched; raw and filtered result counts; images viewed; association or multi-location output; retention settings; audit logs; exported reports; vendor communications; and access policies.
The scale of the network deserves particular attention. A police department may own only a handful of cameras yet have access, through sharing arrangements, to records generated by hundreds or thousands of others. The relevant constitutional capability is what officers could search, not merely what appears on the local agency’s inventory.
Preservation requests should go out early. As of August 2026, Flock states that ordinary LPR image data is deleted after thirty days by default, subject to applicable law and customer arrangements. Alerts, screenshots, exports, case records, or other derived material may persist beyond the ordinary image-retention period. Identify every agency, task force, fusion center, private partner, and vendor that may possess responsive material.
Bottom Line
Chatrie does not make every roadside image a search. It does something more useful: it prevents the government from ending the analysis with “the plate was public.” One camera may record a public fact at one moment. A networked vehicle-location database can do something different. It can reconstruct travel, identify every vehicle near a place, compare vehicles across multiple crime scenes, and generate investigative leads from patterns that no officer could realistically follow at the same scale. The constitutional question therefore cannot be reduced to what one camera happened to see. That is why the label “license plate reader” matters. It makes a networked surveillance system sound like a device performing a single, narrow task. The relevant inquiry is broader: what information did the system collect, how widely was it aggregated and shared, how long was it retained, what databases could officers reach, and what did their query ask the system to reveal?
The plate may be public. The database may contain accumulated observations from cameras the investigating agency does not own. And the query may transform those observations into a history, a pattern, an association, or a suspect. The Fourth Amendment analysis should address the surveillance system that actually produced the evidence—not the narrow label attached to the camera.
The plate is only the index.
Authorities
Chatrie v. United States, 609 U.S. ___ (2026).
Carpenter v. United States, 585 U.S. 296 (2018).
United States v. Jones, 565 U.S. 400, 420 n.3 (2012).
Commonwealth v. McCarthy, 484 Mass. 493, 506–09 (2020).
United States v. Leon, 468 U.S. 897 (1984).
Davis v. United States, 564 U.S. 229 (2011).
Flock Safety materials describing vehicle classification, network sharing, search functionality, audit trails, and default retention.
¹ John C. Ellis, Jr. is a National Coordinating Discovery Attorney for the Administrative Office of the U.S. Courts, Defender Services Office, where he provides litigation support and e-discovery assistance on complex criminal cases to defense teams around the country. Before entering private practice, he spent 13 years as a trial and supervisory attorney with Federal Defenders of San Diego, Inc. He also serves as a digital forensic consultant and expert.
² United States v. Jones, 565 U.S. 400 (2012).
³ Commonwealth v. McCarthy, 484 Mass. 493, 506–09 (2020).