A Case Study of Automated License Plate Recognition Networks
I. Introduction
The Fourth Amendment’s text has not changed since 1791. The technology arrayed against it has. Automated license plate recognition (“ALPR”) networks — exemplified by Flock Safety’s nationwide system, reported by industry and press sources as exceeding 100,000 cameras as of mid-2026, a figure this article has not independently verified — present the doctrine with a fact pattern the founding generation could not have anticipated and that even Carpenter v. United States, 138 S. Ct. 2206 (2018), did not squarely resolve: a privately manufactured, publicly and privately owned, artificial-intelligence-mediated network engineered from inception to function as a law enforcement search tool, indefinitely aggregating the movements of the general public without individualized suspicion.
This article argues that existing doctrine, properly extended rather than merely applied, supports Fourth Amendment scrutiny attaching earlier than the point of query — at the point a surveillance infrastructure is designed and deployed for the primary purpose of enabling law enforcement search. It further argues that this purpose-driven analysis, while doctrinally novel, is not doctrinally unmoored: it synthesizes threads already present in United States v. Jones, 565 U.S. 400 (2012), the special-needs “primary purpose” line of Ferguson v. City of Charleston, 532 U.S. 67 (2001), and City of Indianapolis v. Edmond, 531 U.S. 32 (2000), and Justice Gorsuch’s property-based dissent in Carpenter.
II. The Existing Doctrinal Baseline
A. Public Roads and the Third-Party Doctrine
Under Smith v. Maryland, 442 U.S. 735 (1979), and United States v. Miller, 425 U.S. 435 (1976), information voluntarily exposed to third parties historically carries no reasonable expectation of privacy. Lower courts applying this framework to ALPR include Schmidt v. City of Norfolk, No. 2:24CV621, 2026 WL 207513 (E.D. Va. Jan. 27, 2026) (Davis, J.), granting summary judgment for the city, and State v. Simonson, No. 40732-2-III, 2026 WL 242157 (Wash. Ct. App. Jan. 29, 2026) (unpublished). Both require substantially more qualification than the secondary sources this article originally relied on provided.
Schmidt is a live, precedential district-court decision presently on appeal as Schmidt v. City of Norfolk, No. 26-1227 (4th Cir., docketed Feb. 26, 2026; pending as of this writing). Judge Davis’s 51-page opinion held that Norfolk’s roughly 175-camera, 21-day-retention deployment did not track “the whole of a person’s movements” under Carpenter, but the opinion pointedly did not foreclose future challenges: the court wrote that ALPR surveillance “could become too intrusive” at “some point” as camera density increases, holding only that the system was constitutional “today.” That qualification is directly relevant to this article’s thesis and should not be omitted from any citation to the case.
Simonson requires a more significant correction. It is an unpublished opinion, carrying no precedential value under Washington GR 14.1, and — contrary to characterizations in several secondary sources, including a Flock-published blog post this article’s earlier draft relied on — it is not primarily a Fourth Amendment merits decision. The appeal arose from a trial court’s dismissal of stolen-vehicle charges under CrR 8.3(b) for the State’s failure to preserve ALPR data; the Court of Appeals reversed that dismissal on evidence-spoliation grounds. In the course of assessing prejudice, the panel did address the suppression question, but its merits holding is considerably narrower than “ALPR does not implicate the Fourth Amendment”: it held only that the single photograph actually before the court — one image, from one camera, reviewed by one officer — did not disturb the defendant’s private affairs under Article I, Section 7 of the Washington Constitution, and it expressly distinguished that narrow holding from “the Flock system as a whole,” which it acknowledged “potentially provides broader tracking capabilities.” Notably, the underlying trial court had gone further in the opposite direction, writing that the “expansive technological capabilities of the Flock system, which stores and retains searchable patterns of data… is far different than a single image,” and is “more analogous to how cell towers can track a person’s location and comprehensively chronicle their movements” — language that supports this article’s Part III–IV argument rather than undercuts it. Simonson should be cited, if at all, for its narrow single-photograph holding, not as general authority that networked ALPR querying survives Fourth Amendment scrutiny.
B. Carpenter’s Mosaic Theory
Carpenter carved a narrow but consequential exception: aggregated third-party records that reconstruct the “whole of a person’s movements” implicate the Fourth Amendment notwithstanding voluntary disclosure to the custodian, because such comprehensive retrospective tracking was not within the scope of risk a person assumes by using modern technology. The Court located the constitutional event at the point of government access — the compelled production and examination of Sprint’s records — not at the point Sprint created or retained them. This siting was deliberate: Sprint is not a state actor, and the Fourth Amendment does not reach purely private conduct.
Chatrie v. United States, 609 U.S. ___, No. 25-112 (June 29, 2026), reinforced this framework in the geofence-warrant context. Writing for a 6-3 majority, Justice Kagan held that police conducted a Fourth Amendment search when they obtained Okello Chatrie’s location data from Google via a geofence warrant, reasoning that an individual retains a legitimate expectation of privacy in the whole of his physical movements and that Carpenter’s logic controls notwithstanding the government’s argument that no search occurred at all. Justice Jackson, joined by Justice Sotomayor, concurred; Justice Gorsuch concurred in the judgment, presumably on his positive-law property theory discussed in Part V below; Justice Alito dissented in part, joined by Justice Thomas as to Part I and by Justice Barrett as to several subparts, and Justice Barrett filed a separate dissent. The case arose on certiorari from the Fourth Circuit’s en banc decision, United States v. Chatrie, 136 F.4th 100 (4th Cir. 2025), following the district court’s denial of the motion to suppress, United States v. Chatrie, 590 F. Supp. 3d 901 (E.D. Va. 2022). Chatrie did not resolve the constitutionality of fixed-location ALPR networks or Flock’s national lookup architecture, which involve a materially different collection method than the geofence warrant at issue, but its reaffirmation of Carpenter’s comprehensive-movement-tracking rationale is directly relevant to the query-stage analysis below.
III. Locating the Search Before the Query: A Purpose-Based Theory
A. Jones and the Road Not Fully Taken
Jones held that physically attaching a GPS device to a vehicle was itself a search, independent of whether the resulting data was ever examined — a trespass-based holding. Justice Alito’s concurrence, joined by three others, went further, locating the constitutional harm in the fact of prolonged monitoring rather than in any discrete act of installation or access. Justice Sotomayor’s separate concurrence flagged the distinct danger of government assembly of comprehensive records. Neither concurrence commanded a majority, and Carpenter’s majority opinion did not adopt this framing, choosing instead to locate the search at access.
B. The Primary-Purpose Doctrine as a Closer Analog
A more promising vehicle lies in the Court’s “primary purpose” jurisprudence. In Ferguson, the Court invalidated a hospital drug-testing-and-police-referral program because its primary purpose, assessed at the level of program design, was to generate evidence for law enforcement — not to treat patients. In Edmond, the constitutionality of a highway checkpoint program turned on its primary purpose, evaluated programmatically rather than stop-by-stop. Both cases stand for a proposition directly relevant here: where a program’s animating purpose is law enforcement investigation rather than an independent administrative or commercial function, that purpose is itself constitutionally significant, and courts assess it at the level of the program’s design — not only at the moment of individual application.
Applied to ALPR: Sprint’s cell-site database exists for billing; law enforcement access is incidental and post hoc. An ALPR network’s aggregation and search architecture — Flock’s “National Lookup,” cross-jurisdictional plate matching, and natural-language query tools — exists, by design and by the vendor’s own marketing, for law enforcement investigative search. A database whose primary purpose from inception is enabling government search arguably stands in a different constitutional posture than one whose law enforcement utility is a downstream, unintended byproduct of an unrelated commercial function.
A significant caveat is required here. Ferguson and Edmond do not hold that a program’s purpose determines whether a search has occurred at all; both cases assume a search is already conceded (a urine test, a vehicular stop) and use primary purpose only to determine what level of justification — ordinary probable cause versus a relaxed “special needs” standard — that already-conceded search requires. Extending that purpose-based reasoning to the antecedent question of whether a search has occurred at the design stage, before any individual application, is an analogical extension this article proposes, not a direct application of existing holdings. A rigorous referee should expect this move to be defended on its own terms rather than presented as settled by Ferguson or Edmond themselves.
IV. The State-Actor Question for Privately Owned Nodes
A substantial share of ALPR nodes — HOA, apartment complex, and private business cameras — are privately owned, raising a threshold question: can Fourth Amendment scrutiny attach to data collected by a private party at all?
A. The Jackson/Rendell-Baker Bar
Jackson v. Metropolitan Edison Co., 419 U.S. 345 (1974), and Rendell-Baker v. Kohn, 457 U.S. 830 (1982), establish that heavy government regulation or near-total government funding does not, standing alone, convert a private actor into a state actor. A private entity’s financial dependence on government contracts — even overwhelming dependence — is insufficient.
B. Reviving Joint Action Through Purpose
Skinner v. Railway Labor Executives’ Ass’n, 489 U.S. 602 (1989), permits a finding of state action where the government significantly encourages specific private conduct, or where a private party’s conduct is undertaken to serve a governmental function rather than an independent private one. This is where the primary-purpose analysis in Part III does double duty: a private camera operator whose device was purchased, installed, and configured specifically to feed a law-enforcement-searchable network — pursuant to a commercial product expressly marketed and architected for that governmental function — presents a materially different fact pattern from Rendell-Baker’s government-funded but functionally independent private school. The purpose-built character of the infrastructure is evidence of joint action that mere payment, without more, cannot supply.
V. An Alternative Path: Property and Custodial Obligation
Justice Gorsuch’s dissent in Carpenter proposed an approach the majority declined to adopt: that positive law — including bailment principles — may leave the data subject with a property interest in records physically held by a third-party custodian, such that Fourth Amendment protection does not evaporate merely because another party possesses the data. Under this framework, a government agency’s ALPR data remains, in a meaningful sense, the agency’s property notwithstanding Flock’s custodial possession and architecture; the vendor’s purpose-built search capability becomes a custodial arrangement structured to grant frictionless access to the beneficial owner — the government — a relationship warranting closer scrutiny than an unrelated business incidentally holding data of interest to police. No majority opinion has adopted this theory, but it offers a textually and historically grounded alternative that does not require resolving the state-action question at all for government-owned camera deployments.
VI. An Aggravating Factor: Biometric Convergence
A separate and underexplored risk concerns data-stream convergence. Current ALPR products, including Flock’s, do not store facial identity data — indeed, face-detection routines are used affirmatively to blur faces in stored imagery, a privacy-protective design choice. However, vendors in this space, including Flock’s own broader product suite, separately market facial-recognition-capable camera systems to the same public and private customer base operating on the same underlying platform. Should vehicle-movement data and facial-identity data ever be merged, cross-referenced, or made jointly queryable — whether by policy change, product integration, or unauthorized access of the kind already documented in several 2026 municipal audits — the “vehicle characteristics only, not identity” defense that currently forecloses stronger privacy claims would no longer describe the system in operation. Courts and legislatures assessing ALPR constitutionality on the current record should not treat the absence of biometric identification as a permanent structural feature; it is a policy choice within a single vendor’s control, on a platform architecturally capable of the alternative.
VII. The Overbreadth Objection and a Limiting Principle
Any design-purpose theory of this kind invites an obvious objection: nearly every law enforcement database — NCIC, state DMV records, fingerprint archives, booking photo repositories — is, in some sense, “purpose-built” to aid government search. If purpose alone triggers Fourth Amendment scrutiny at creation, the theory threatens to sweep in databases whose constitutionality has never seriously been questioned, and to do so without a workable stopping point.
The distinction this article proposes is not purpose alone but the combination of purpose and comprehensiveness of the underlying liberty interest at stake — specifically, aggregated, retrospective tracking of physical movement over time, the precise interest Carpenter identified as categorically different from discrete, static records. Fingerprint or booking-photo databases store static identifying information collected in connection with a specific, individualized law enforcement encounter (an arrest, a booking); they are not populated by continuously and indiscriminately monitoring the public at large. ALPR networks, by contrast, are populated by suspicionless, continuous, general-population surveillance whose aggregate product is a movement history indistinguishable in kind from the cell-site data at issue in Carpenter. The proposed design-stage scrutiny is limited to databases sharing both features — investigative purpose and suspicionless, continuous collection producing a comprehensive movement or associational record — and would not extend to conventional law enforcement recordkeeping populated through individualized, suspicion-based process. Whether this line can be administered consistently by courts, rather than merely stated, is a fair question this article does not fully resolve and flags as the theory’s principal vulnerability.
VIII. A Proposed Framework
Synthesizing the foregoing, this article proposes that Fourth Amendment analysis of purpose-built surveillance networks should proceed on three independent tracks, any one of which may suffice:
Query-stage scrutiny (settled ground): retrospective, cross-time, cross-jurisdiction queries capable of reconstructing a comprehensive movement history are searches under Carpenter and require a warrant, regardless of camera ownership. Narrow, real-time, single-target queries analogous to a traditional BOLO remain outside this rule.
Design-stage scrutiny (novel but doctrinally grounded): where a surveillance infrastructure’s primary purpose, assessed programmatically under Ferguson/Edmond, is to enable law enforcement search, constitutional scrutiny should attach to the government’s deployment and configuration of that infrastructure independent of any specific query — at minimum for government-owned nodes, and, on a joint-action theory, potentially for privately-owned nodes purpose-built for that function.
Custodial/property scrutiny (alternative, unadopted but available): government data held by a private vendor architected to provide frictionless law-enforcement access may be treated as remaining the government’s property for Fourth Amendment purposes under a bailment theory, obviating the need to resolve state action for the vendor’s conduct at all.
IX. Conclusion
No court has yet adopted the full synthesis proposed here. The query-based Carpenter theory remains the most litigated and best-supported avenue and should be regarded as the primary vehicle for near-term challenges. But the design-purpose and custodial-property theories are not speculative additions; they are extensions of reasoning already present, in fragments, across the Court’s own opinions. A network built, marketed, and sold expressly as a law enforcement search tool — regardless of who owns any individual node, and regardless of whether any individual query has yet occurred — sits closer to the general warrants the Fourth Amendment’s framers set out to foreclose than to the individualized, suspicion-based searches the Amendment was principally drafted to regulate. Doctrine has not yet caught up to that architecture. It should.
This article synthesizes analysis developed through Socratic dialogue and is offered as a framework for further legal research, not as a summary of settled law. Case citations reflect the state of authority as of July 2026 and, as to Schmidt v. City of Norfolk*,* State v. Simonson*, and* Chatrie v. United States*, have been verified against primary-source dockets and opinions rather than secondary reporting alone.* Schmidt remains on appeal to the Fourth Circuit and should not be cited as final; the correction of Simonson*’s scope in Part II.A above should be treated as authoritative over any prior secondary characterization of that case, including this article’s own initial draft.*
