Florida Is Pulling Plate Readers From State Roads. The Evidence Shows Why

Automated license-plate readers can help recover stolen vehicles. Erroneous alerts, unauthorized searches and uncontrolled data sharing help explain Florida’s abrupt retreat from the technology.

Florida has ordered automated license-plate readers removed from state highway rights-of-way, placing the state at the center of a national debate over whether the technology’s public-safety benefits justify its financial and civil-liberties costs.

On Aug. 31, the Florida Department of Transportation revoked permits previously issued to law-enforcement agencies and gave them 30 days to remove affected equipment. The directive applies to automated license-plate readers located within state highway rights-of-way, including cameras manufactured by Flock Safety and other companies.

It does not automatically prohibit readers positioned along county roads, municipal streets or private property.

Some Florida agencies are going further. Miami-Dade suspended its entire license-plate-reader program, while sheriffs in Jacksonville and several other jurisdictions announced broader shutdowns or removals. Major Tampa Bay agencies, however, indicated that they would comply with the state-road directive without necessarily abandoning readers installed elsewhere.

The result is not a complete statewide ban. It is a developing patchwork in which cameras may be removed from one road but remain operational nearby.

Gov. Ron DeSantis framed the decision as an effort to prevent Florida from becoming what he called a “digital AI surveillance state.” FDOT cited the rapid expansion of automated readers, reported misuse, data-privacy concerns and broader surveillance practices.

Those concerns are not hypothetical. Florida misconduct investigations, government audits and disclosures involving unauthorized access have documented ways automated plate-reader systems can expose innocent motorists to police intervention or allow officials to search vehicle-location records for personal reasons.

The evidence also complicates any claim that the cameras provide no public benefit. Automated readers have helped police recover stolen vehicles, locate missing people and identify vehicles associated with serious crimes.

The relevant question is not whether the technology ever works. It plainly does.

The question confronting Florida is whether those successes are substantial enough—and sufficiently documented—to justify a system that records the movements of millions of people who are not suspected of crimes.

A system that records ordinary movement

Flock Safety’s automated license-plate readers photograph passing vehicles and convert their plates into searchable information.

The resulting record can include a vehicle’s plate number, make, model, color, location and time of observation. Depending on an agency’s settings and agreements, authorized users may search the information later or share access with other law-enforcement organizations.

Unlike a conventional surveillance camera reviewed after a particular incident, a network of automated readers creates records of vehicles before police have any reason to suspect their occupants of wrongdoing.

That distinction matters.

A searchable vehicle database can help investigators locate a kidnapping suspect or find a stolen car. The same capability can be used to follow an ex-partner, identify people attending a political demonstration or conduct searches unrelated to legitimate police work.

Flock says searches are recorded in audit logs and that users must supply a reason for accessing the system. Its policy also advises officers to verify computer-generated information before taking enforcement action.

Recent cases demonstrate that logging searches does not necessarily prevent unauthorized searches from occurring.

Florida cases illustrate the risk of officer misuse

In Brevard County, former sheriff’s deputy Michael Fultz was accused of improperly using the agency’s Flock system to search for his former girlfriend’s license plate.

According to an administrative investigation reported by News 6, an audit determined that Fultz conducted the search before his shift and entered “investigation” as the justification. Investigators reportedly found no legitimate law-enforcement purpose for the search.

Fultz resigned in May as the sheriff’s office prepared to fire him following an investigation into multiple forms of alleged misconduct. He had been recognized as the agency’s Deputy of the Year only months earlier.

The incident was not an example of a camera misreading a license plate. It was an example of an authorized user allegedly employing a police tracking system for a personal purpose.

A separate case surfaced in Sarasota in August.

Former Sarasota police officer Cory Waiters was fired and arrested after investigators accused him of misusing law-enforcement databases, including Flock and Florida’s Driver and Vehicle Information Database.

Police alleged that Waiters used Flock to conduct 335 searches involving the same license plate between May 20 and July 16. He faces charges that include official misconduct and an offense involving the misuse of a computer system.

Those charges remain allegations unless proven in court.

Together, the cases expose the limitations of relying principally on audit logs and required justification fields. A user conducting an improper search can enter a vague explanation such as “investigation.” Unless an agency actively compares searches with case files and examines unusual patterns, misuse may remain undiscovered.

A nationwide review by The Washington Post identified at least 69 police officials who had reportedly been accused, charged or convicted of misusing Flock or other license-plate-reader systems. In at least 15 cases, the potential misuse was first identified by someone outside the responsible police department.

Flock has announced that an auditing feature intended to identify unusual searches will become mandatory for law-enforcement customers. That could improve oversight, but automated anomaly detection still depends on whether the system recognizes suspicious behavior and whether an agency acts on the resulting warning.

A false alert does not always mean the camera misread a plate

Discussions of automated license-plate readers frequently collapse several different failures into the phrase “false positive.”

Sometimes the character-recognition software misreads a license plate. Similar letters and numbers, obscured characters, specialty plates and incorrect identification of the issuing state can produce inaccurate matches.

But a camera can read every character correctly and still initiate an encounter with an innocent person.

A recovered vehicle may not have been removed promptly from a stolen-vehicle database. An officer may enter incomplete plate information. A plate may have been stolen, duplicated or attached to another vehicle. The vehicle may match an investigative description while its current driver has no connection to the suspected crime.

To someone ordered from a vehicle during a high-risk stop, the source of the mistake may make little practical difference.

The distinction is nevertheless important for accountability. Describing every bad stop as an artificial-intelligence error can allow police agencies, database administrators and individual officers to avoid responsibility for inaccurate records or inadequate verification.

An automated reader, a police hotlist and an officer’s decision operate as one system. A failure at any point can place an innocent motorist in a dangerous encounter.

Los Angeles provides a rare comparison

A 2026 review by the Los Angeles Police Department’s Office of the Inspector General provides one of the clearest public examinations of ALPR benefits and failures.

During a two-month period covering August and September 2025, the department’s ALPR systems contributed to the recovery of 337 stolen vehicles and 74 arrests. Those figures demonstrate that license-plate readers can generate meaningful investigative results.

The same review found 161 stops involving alerts that matched the plate entered into the system but vehicles subsequently determined not to be stolen. The inspector general attributed the problem primarily to inaccurate or outdated information in law-enforcement databases.

Those findings require an important qualification: Los Angeles used ALPR technology involving Flock, Axon and Motorola. The 161 incidents cannot responsibly be described as errors caused exclusively by Flock cameras.

They instead demonstrate a broader systems problem.

The camera may correctly read the plate while the database falsely tells an officer that the vehicle is stolen. Unless that information is independently verified, a technically accurate camera reading can still result in an unnecessary detention.

The inspector general recommended stronger oversight, standardized documentation of ALPR-related stops and suspension of additional deployments until the city received more public input. Los Angeles subsequently paused its relationship with Flock while addressing concerns involving data ownership, security and sharing.

A system that contributed to 337 vehicle recoveries while also producing 161 unnecessary stolen-vehicle stops cannot honestly be summarized as either an unqualified success or a complete failure.

It demonstrates why agencies must report both outcomes.

Evidence of effectiveness remains incomplete

The strongest emerging evidence for Flock concerns motor-vehicle theft.

An August 2026 working paper examined the staggered deployment of fixed Flock readers across 216 American law-enforcement agencies using federal crime data covering 2017 through 2023.

Researchers estimated that motor-vehicle theft fell approximately 11% following deployment and reported improved clearance outcomes.

That is meaningful evidence in the technology’s favor, but it is not the final word. The study was released as a working paper, meaning its findings should not be treated as settled peer-reviewed research. Its most direct conclusions concern vehicle theft—not crime generally.

Earlier ALPR research produced mixed results. A 2025 evaluation in Atlantic City reported reductions in shootings, motor-vehicle theft and property crime but did not find a broader reduction in violent crime. Researchers have also noted the difficulty of separating a camera’s effects from changes in police staffing, enforcement strategies and underlying crime trends.

Flock publishes its own effectiveness estimates and customer surveys indicating that its systems contribute to large numbers of investigations. Those claims deserve consideration, but vendor-sponsored research, participating-agency surveys and selected success stories are not substitutes for independent evaluation.

The defensible conclusion is narrower: automated readers can help locate particular vehicles and may improve outcomes involving vehicle theft.

Evidence supporting broader claims—that they substantially reduce crime generally, make entire communities safer or provide benefits proportionate to their expanding scale—is considerably less certain.

Data sharing creates another category of risk

Even if every license plate is read correctly and every local officer behaves appropriately, an agency may lose control over where its information travels.

Audits in California found that vehicle information collected by some local departments became accessible to federal or out-of-state agencies despite state restrictions or local policies.

Ventura County officials said outside agencies searched locally collected ALPR information hundreds of thousands of times during a two-month period without the department’s knowledge or approval.

Flock suggested that a user, company employee or software problem could have reactivated a national lookup function. The sheriff’s office said its investigation found that nobody within the agency activated it.

Mountain View and Oxnard also disclosed unintended or unauthorized access involving federal or out-of-state agencies. Flock has said it implemented additional protections following those disclosures.

These incidents complicate one of the most common assurances offered at local approval meetings: that the purchasing agency owns and completely controls its information.

Nominal ownership provides little protection if software settings, system integrations or contractual provisions permit access that the agency did not authorize—or did not know was occurring.

Florida’s order leaves a local patchwork

Florida’s directive reduces the physical presence of automated readers without necessarily dismantling the larger surveillance network behind them.

A local agency may remove cameras situated within state highway rights-of-way while continuing to operate readers on municipal or county property. Agencies without active cameras may also retain the ability to request information collected in neighboring jurisdictions.

The state will eventually need to decide whether removing readers from state roads is sufficient or whether Florida needs uniform rules governing:

• Permissible reasons for searching vehicle records
• Warrants for historical location searches
• Data-retention periods
• Federal and out-of-state access
• Required audits of officer activity
• Verification before an enforcement stop
• Reporting of alerts, errors, arrests and prosecutions
• Penalties for unauthorized searches
• Replacement of Flock cameras with competing systems

Without broader rules, agencies could relocate cameras away from state highways or replace one vendor with another while preserving the same underlying surveillance capability.

The public-policy question is not limited to Flock. It concerns any system capable of recording and searching vehicle movements at a comparable scale.

Pasco’s drone proposal remains relevant—but it is different

Pasco County’s proposed $9 million drone-as-first-responder network should not be presented as an expansion of Flock license-plate readers.

The county proposal concerns autonomous police drones. The preliminary funding memorandum calls for 21 weather-rated docking stations and 22 aircraft, but it does not identify the eventual vendor.

The proposal remains relevant as an example of a Florida government considering another form of permanent, automated law-enforcement infrastructure while the state retreats from roadside plate surveillance.

The Pasco system raises many of the same oversight questions: what information will be collected, how long it will be retained, which agencies will have access, whether drones will be dispatched using automated vehicle alerts and how the county will measure benefits against costs and mistaken deployments.

Florida’s reversal on plate readers makes those questions more urgent.

Local governments should establish safeguards and performance measures before deploying a new surveillance system rather than attempting to impose accountability after problems emerge.

The missing denominator

Police departments commonly announce that Flock helped recover a stolen vehicle, locate a missing person or arrest a violent suspect. Those successes are real and newsworthy.

They are not sufficient to evaluate the system.

A meaningful assessment would disclose:

• The total number of plates scanned
• The number of automated alerts generated
• How many alerts were verified before a stop
• How many alerts relied on inaccurate or outdated information
• The number of resulting stops and arrests
• The number of charges filed and convictions obtained
• The number of stolen vehicles recovered
• The number of innocent motorists subjected to high-risk stops
• How many searches lacked a documented case number
• How many users were investigated or disciplined for improper access
• Which outside agencies searched locally collected information
• The system’s full purchase and recurring operating costs

Without those figures, the public receives success stories without a denominator.

An agency can count every investigative “assist” while failing to count searches that produced nothing, innocent people stopped, charges dismissed, misconduct investigations or cases that never resulted in prosecution.

That is not enough information to establish that population-scale surveillance produces a net public benefit.

A tool whose value depends on restraint

Flock cameras do not arrest people. They do not update stolen-vehicle databases, verify alerts, supervise officers or determine which outside agencies should receive access.

People and institutions make those decisions.

That does not absolve the technology company. Flock designs the network, establishes available safeguards, writes standard contractual terms and markets the system’s capabilities to governments.

But the public-policy question is larger than whether a camera correctly reads a license plate.

The relevant question is whether the entire system—camera, database, officer, department policy and multistate network—produces more public benefit than harm.

Current evidence offers a mixed answer.

Automated readers can help recover stolen vehicles and identify suspects. They can also expose innocent motorists to dangerous police stops, provide abusive officials with an efficient tracking tool and make locally collected information available far beyond the jurisdiction that installed the camera.

Florida’s order represents a significant retreat from automated roadside surveillance, but it does not settle the debate.

As agencies decide whether to discontinue their programs, remove only cameras covered by FDOT’s directive or relocate readers to other roads, residents deserve more than isolated success stories.

Agencies should disclose how frequently their systems work, how frequently they fail, who has searched the information and what happens when the rules are broken.

If an agency cannot provide those answers, it cannot credibly claim to know whether retaining the cameras makes the public safer.

Sources & Editorial Note

This is a news-analysis article. Its central assessment—that the demonstrated benefits of automated license-plate readers may not outweigh their financial, privacy and civil-liberties costs—is analysis rather than a settled empirical finding. Evidence reviewed for this article indicates that ALPR systems can assist with stolen-vehicle recoveries and some criminal investigations. No comprehensive nationwide study reviewed establishes that these systems are either universally beneficial or universally counterproductive.

Florida’s action is based on the Florida Department of Transportation’s Aug. 31, 2026, revocation of permits for automated license-plate readers. The directive applies to readers within state highway rights-of-way and is not a complete statewide prohibition on every ALPR. Subsequent agency reactions were reported by Axios Tampa Bay, CBS Miami and WUFT.

Findings concerning Los Angeles are based on the Los Angeles Police Department Office of the Inspector General’s review of LAPD’s automated license-plate-reader system and CBS Los Angeles reporting. LAPD used ALPR technology involving multiple vendors, so its erroneous-alert figures should not be attributed exclusively to Flock Safety.

Research concerning motor-vehicle theft is based on the August 2026 working paper examining fixed Flock deployments across 216 agencies. The paper had working-paper status when this article was prepared and should not be described as settled peer-reviewed research.

The Brevard County findings involving Michael Fultz are based on an administrative investigation described by News 6. The Sarasota allegations involving Cory Waiters are based on police statements and court records summarized by Patch. Criminal allegations against Waiters have not been proven unless and until established in court.

National reporting about alleged misuse, public opposition and Flock’s announced safeguards is drawn from The Washington Post and the Associated Press. Flock’s description of its policies and safeguards is available in the company’s license-plate-reader usage and privacy policy.

Details of Pasco County’s drone proposal are based on the county agenda memorandum for Application Case 32774. The preliminary document recommends a $9 million allocation but does not, by itself, identify the eventual vendor or establish the final terms of a procurement.

Flock Safety and the Florida agencies discussed should be offered an opportunity to respond before publication. Any response received should be incorporated into the article or the article should state that the organization did not respond or declined to comment.