Cops Are Using These Cameras to SPY on Their Exes

August 23, 2026

License plates are public by design, but when millions of scans are aggregated, retained, and searched at scale, they become a map of people’s lives—useful for policing, yes, but also fraught with opportunities for error, abuse, and mission creep.

At a Glance

Automated license plate readers (ALPRs) like Flock Safety’s deliver real policing value, but the database they create is overwhelmingly composed of scans unrelated to crime.
Documented misuse by sworn officers shows policy and audit logs alone do not prevent improper searches; stronger technical guardrails and external oversight are needed.
Flock is rolling out seven-day default retention, evidence-preservation tools, and anomaly detection with lockouts; the efficacy of these controls hinges on implementation and independent verification.
The core policy choice isn’t whether plates are public—it’s whether mass, persistent collection and sharing should be limited by time, purpose, and governance to protect civil liberties.

How ALPRs Work—and Why Scale Changes the Privacy Equation

Automated license plate readers are simple systems with complex consequences. Fixed or mobile cameras capture plate numbers and timestamps; software translates images into text (optical character recognition), compares results against “hot lists,” and stores plate, time, and location for later queries. A single scan is banal. A persistent network is not. At scale, the dataset describes patterns—work commutes, clinic visits, religious attendance—that the Supreme Judicial Court of Massachusetts and other courts have flagged as constitutionally sensitive when surveillance becomes widespread and persistent, even if any one observation is public. The legal conversation has shifted from whether a plate is visible to whether aggregating it across time triggers heightened protections.

The base-rate reality makes the governance problem unavoidable: the vast majority of scans are of cars connected to no wrongdoing. The Brennan Center has emphasized that the percentage of “hits” is generally far below one percent, which means most of the database is information about ordinary, lawful movement. If retention and sharing are generous, a city quietly amasses a longitudinal location registry of its residents; if controls are tight and time-limited, the same cameras function more like a transient alerting tool, closer to a modernized “be on the lookout” notice than a dragnet archive. That design choice is policy, not inevitability.

What Recent Misuse Incidents Actually Prove

Debate over hypothetical harms can feel abstract; the misuse record is not. In Reynoldsburg, Ohio, public records show an internal police inquiry concluded that one officer ran searches on the name of a current or former partner 46 times since 2025 while on light duty or leave, and another officer conducted over fifty searches on a single plate in one year. Those are not edge cases in a lab—they are audit entries pulled from a production system, precisely the sort of “non-investigative” lookups that internal policy is supposed to deter. The point isn’t that all officers misuse tools; it’s that any system that can be queried for personal reasons eventually will be, unless the friction to do so is high and detection is automatic and external. Audit logs help after the fact; they do little in the moment unless paired with real-time anomaly detection and consequences.

Reformers often cite these cases to argue for shorter default retention and stronger purpose binding. They are right on the mechanism: less data and narrower permissible queries reduce the blast radius of abuse. Misuse also undercuts the most common reassurance—“every search is recorded”—because logging is necessary but insufficient. A seatbelt is good; you still need brakes.

Flock’s Guardrails: What’s Promised and What Matters

Flock Safety emphasizes that it does not use facial recognition, does not sell customer data, and that access is controlled by local agencies with role-based permissions. The company says every search is tied to a user account and must be for a specific investigation, recorded for supervisory review. These are sensible baselines, not end states. In response to mounting criticism, Flock is lowering its default retention from 30 days to seven, adding an “Evidence Mode” to preserve specific records for active cases, and requiring case numbers on searches, with automatic review and proactive lockouts for abnormal activity. If rigorously implemented, these steps directly target the failure modes revealed by real incidents: excessive retention, ad hoc queries, and post hoc accountability that arrives too late.

The real test is operational, not rhetorical. Does the system hard-stop an officer who can’t supply a valid case number? Are anomaly thresholds calibrated to flag serial queries on the same plate or searches by off-duty personnel—and do they trigger immediate suspension pending review? Can local administrators quietly relax settings, or are certain safeguards pinned by contract and visible to the public via transparency portals? “Trust center” pages are not oversight; independent audits and enforceable procurement terms are.

Effectiveness Claims and the Public-Safety Ledger

Proponents argue the technology helps solve crime and locate missing people, pointing to headline figures of crimes solved and recoveries. The underlying story is straightforward: alerts match wanted plates in real time, and retrospective searches can place a suspect vehicle near a scene. Those mechanisms are real; so are false reads, stale hot lists, and downstream police action on bad data. The utilitarian case isn’t null—it is conditional. ALPRs are most defensible when used for narrow, time-bound tasks like responding to a fresh 911 call or a child abduction, where the value of speed is high and the retention need is short. The further use drifts toward building generalizable movement histories, the weaker the safety dividend relative to the privacy cost, especially given the base rate of innocence in the dataset.

Communities do not have to accept an all-or-nothing trade. Purpose limitation, short retention, strict sharing controls, and externally visible auditing can preserve much of the acute policing benefit while degrading the surveillance value of the archive. That is architecture, not aspiration.

Law, Policy, and the Line Between Public and Persistent

Courts across jurisdictions have reached varied but converging views: observing a plate on a public road is not novel, yet widespread, persistent tracking can implicate constitutional interests. Massachusetts’s high court, for example, has recognized that “widespread use” changes the analysis, while other decisions have upheld specific programs under particular facts. Policy is moving in tandem, with legal scholars and bar associations urging limits on retention, purpose binding, and access. This is a familiar trajectory in surveillance law: what police can do one officer at a time without a warrant becomes qualitatively different when automated, aggregated, and queryable at will. The doctrine is catching up to the engineering.

For city councils and procurement officers, the legal variance argues for conservative design: codify seven-day or shorter retention, mandate case-numbered searches with cryptographically signed logs, define disclosable quarterly transparency reports, and require independent audits with public summaries. Contracts should hard-code non-negotiable guardrails, including prohibitions on third-party sharing beyond enumerated partners, with violation penalties that exceed the cost of doing business.

Flock cameras are NOT good. Mass surveillance has just been placed on steroids with the implementation of Flock Camera technology! All the pieces have to be in place for the next Plandemic!

'We're Not Big Brother': Flock CEO Unveils New Privacy Guardrails After Backlash…

— Charlie (@MAGACharlie2024) August 19, 2026

A Governance Blueprint That Meets the Moment

Done right, ALPRs operate more like a just-in-time alerting layer than a standing dossier. The blueprint is straightforward: default retention measured in days, not weeks; active preservation tied to a case number and supervisory approval; immutable audit trails reviewed by an external inspector general or civilian board; automatic lockouts for anomalous use; public, quarterly disclosure of usage metrics, hit rates, retention exceptions, and cross-agency sharing; and sunset clauses that force reevaluation. Pair these with training that treats misuse as termination-grade misconduct, and the incentive structure changes. Communities get the immediate signal when a stolen vehicle pings; bad actors get clipped at the log-in screen.