Flock Cameras and Automated License Plate Recognition: Surveillance, Public Safety, and Oversight

confidence: high · 1 sources

What the evidence says

The sources on this page are a single descriptive paper on Flock Safety's ALPR system (Jeba Sania, 2026). It has no outcome data. It reports no crime rates, clearance rates, error rates, or counts of misuse incidents. Every finding below is a description of how the system is built and overseen, and each rests on one source.

On scale, the paper reports local ALPR cameras in over 6,000 cities. It also reports integration with state and national databases, including the FBI's National Crime Information Center (NCIC) database. The paper says this increases surveillance infrastructure. It gives no figures for cities added or data volume. It also says many national law enforcement databases are routinely out of date or erroneous. It states that this increases the risk of unwarranted harassment. It does not quantify the error rates or document any incidents.

On oversight, the paper describes two gaps. Customers such as businesses, HOAs, and private citizens own the locally collected data. According to the paper, these users receive no legal or ethical training and generally face weaker oversight standards than law enforcement. Law enforcement users must log a purpose when they search the database, but investigations found that queries often use vague language. The paper says this makes audits of system use ineffective. The paper also states that Flock Safety cannot detect user-initiated abuse. The claims do not say whether the company tried to build such detection.

On structure, the paper reports that informal data-sharing agreements between private users and law enforcement give federal agencies workarounds around state ALPR laws and legal constraints. It names no specific laws or constraints. It also reports that decentralized deployment and network effects make the system harder to dismantle collectively without centralized decision-making. Inconsistent user-led deployments can lead to inconsistent privacy safeguards. No claim states which safeguards vary or by how much.

The page has no measured evidence on public safety outcomes such as crime reduction, case clearance, or stolen-vehicle recovery. It also has no evidence on wrongful stops, or on whether any oversight rule changes misuse rates. These outcomes are what people most often expect a review of ALPR to cover, and the sources here do not supply them.

Where the evidence splits

No contradictions in the current sources.

Thresholds and dose-response

No threshold established in the current sources.

Open empirical questions

What effect does ALPR deployment have on crime, case clearance, or vehicle recovery? The sources here report none of these outcomes.
What share of ALPR database hits rest on outdated or erroneous records, and how many lead to unwarranted stops or harassment?
How often do vague search-purpose entries appear in audit logs, and how often does misuse go undetected?
Which state laws are circumvented through private-to-police data sharing, and how often does it occur?

Findings

ALPR deployment in over 6,000 US cities, integrated with national databases, increases surveillance infrastructure

confidence — direction: highconfidence — magnitude: mediumconsensus: thin — one descriptive source, no dissent in the claims

mechanism: local cameras feed a networked system that connects to state and national databases such as the FBI's NCIC database
effect: increases surveillance infrastructure capacity; no magnitude given beyond "over 6,000 cities"
conditions / moderators: not established

Jeba Sania (2026), arXiv (Cornell University) · doi:10.48550/arxiv.2606.17286 · https://arxiv.org/pdf/2606.17286 (retrieved 2026-09-30)

Outdated or erroneous integrated databases increase the risk of unwarranted harassment

confidence — direction: highconfidence — magnitude: lowconsensus: thin — one source, stated as a reported risk

mechanism: ALPR hits are checked against national law enforcement databases that are reported to be routinely out of date or erroneous
effect: increases risk of unwarranted harassment; no error rates or incident counts given
conditions / moderators: not established

Jeba Sania (2026), arXiv (Cornell University) · doi:10.48550/arxiv.2606.17286 · https://arxiv.org/pdf/2606.17286 (retrieved 2026-09-30)

Private, non-state operators face weaker oversight than law enforcement

confidence — direction: highconfidence — magnitude: mediumconsensus: thin — one source

mechanism: businesses, HOAs, and private citizens own the locally collected data and receive no legal or ethical training
effect: decreases oversight rigor for non-state actors; no comparative metric given
conditions / moderators: applies to non-state customers of the system; not established for other operator types

Jeba Sania (2026), arXiv (Cornell University) · doi:10.48550/arxiv.2606.17286 · https://arxiv.org/pdf/2606.17286 (retrieved 2026-09-30)

Informal private-to-police data sharing allows federal agencies to work around state ALPR laws

confidence — direction: highconfidence — magnitude: mediumconsensus: thin — one source

mechanism: data-sharing agreements between private citizens and law enforcement provide federal agencies with workarounds to avoid breaking state laws on ALPR data and legal constraints
effect: increases circumvention of legal constraints; no examples or counts given
conditions / moderators: not established

Jeba Sania (2026), arXiv (Cornell University) · doi:10.48550/arxiv.2606.17286 · https://arxiv.org/pdf/2606.17286 (retrieved 2026-09-30)

Vague search-purpose entries decrease the usefulness of audit logs

confidence — direction: highconfidence — magnitude: mediumconsensus: thin — one source

mechanism: law enforcement users must log a purpose for each database search, but queries often use vague language, which the paper says renders audits ineffective
effect: decreases audit effectiveness; no audit failure rate given
conditions / moderators: applies to law enforcement searches that use vague purpose text; the paper cites investigations but gives no method or rate

Jeba Sania (2026), arXiv (Cornell University) · doi:10.48550/arxiv.2606.17286 · https://arxiv.org/pdf/2606.17286 (retrieved 2026-09-30)

Flock Safety cannot detect user-initiated misuse of the system

confidence — direction: highconfidence — magnitude: highconsensus: thin — one source

mechanism: the paper states the company lacks the ability to detect user-initiated abuse, so misuse is not flagged automatically
effect: no automated abuse detection exists; magnitude not quantified
conditions / moderators: the claims do not say whether detection was tried and failed or never attempted

Jeba Sania (2026), arXiv (Cornell University) · doi:10.48550/arxiv.2606.17286 · https://arxiv.org/pdf/2606.17286 (retrieved 2026-09-30)

Decentralized, user-led deployment decreases regulatory feasibility and increases variation in safeguards

confidence — direction: highconfidence — magnitude: medium for regulatory feasibility, low for safeguard variationconsensus: thin — one source

mechanism: the system is decentralized and has network effects, so without centralized decision-making it is harder to dismantle collectively; inconsistent user-led deployments can lead to inconsistent privacy safeguards
effect: decreases regulatory feasibility (no magnitude); increases variation in safeguard implementation (no magnitude)
conditions / moderators: not established; no claim names which safeguards vary

Jeba Sania (2026), arXiv (Cornell University) · doi:10.48550/arxiv.2606.17286 · https://arxiv.org/pdf/2606.17286 (retrieved 2026-09-30)