It’s been a tough summer for automatic license plate readers (ALPRs). Earlier this month, the Washington Post reported that at least 50 police officers have been charged with or accused of misusing plate-reader networks, including for stalking wives, ex-girlfriends, and other women. One small-town Georgia police chief ran his ex-girlfriend’s plates roughly 600 times. California legislators took up the technology in hearings titled “Don’t Flock me,” a jab at Flock, the Atlanta company whose ALPRs now scan roads in more than 6,000 American communities. A backlash that began among left-leaning civil libertarians has spread to the populist Right, and city councils now face packed hearings on whether to keep the cameras.
The debate has become more heat than light. Both sides fixate on anecdotes, with supporters pointing to abducted children recovered after a timely plate hit, while critics list stories of rogue officers and plate data shared far beyond the agencies that collected this information. Neither answers a question that policymakers face: Does the technology produce enough public safety to justify its costs, in dollars and in privacy?
Finally, a reason to check your email.
Sign up for our free newsletter today.
While we can’t answer this question definitively, our recent working paper should help. We’ve published what we believe to be the first rigorous analysis of how ALPRs affect crime, finding that they are associated with an 11 percent reduction in motor vehicle thefts. That’s not the end of the discussion, but it should help push the discourse in a more nuanced direction.
Thus far, the research on ALPRs has been frustratingly meager. The strongest studies are two randomized experiments, both more than a decade old, and both assessing a different technology: camera readers mounted on patrol cars. In the first, Cynthia Lum and colleagues at George Mason University assigned camera-equipped patrols to auto-theft hot spots and found no detectable effect on theft. The second, in Arizona, produced more stolen-vehicle recoveries and arrests but no decline in theft itself.
Both studies tell us little about the modern version of ALPRs. Fixed cameras, solar-powered and sold by subscription, transformed plate-reading from a tool that a limited number of officers drove around with into stationary infrastructure that watches a road continuously.
Beyond those older studies, the ALPR literature relies on correlational evidence, limited in what it can tell us about how the technology causally affects crime. Comparisons like those published by Flock itself reveal as much about which cities have crime problems as about what the cameras can do for them. Reviewing the evidence recently in The Atlantic, the Manhattan Institute’s Charles Fain Lehman argued that none of this evidence delivers causal proof. The best research is on stationary surveillance cameras in Medellín, Stockholm, Montevideo, and China—neither American implementations nor necessarily the same as ALPRs.
That was a then-accurate reading of the literature. But our new study constitutes what is, to our knowledge, the first evaluation of fixed plate-reader networks.
Our study is based on one major ALPR vendor’s installation records, which identify the month that 216 law enforcement agencies activated their first camera. We combined those data with the monthly crime data those agencies reported to the FBI’s incident-based system. Another 1,800 agencies in the same population range served as our control group.
Our study focuses on motor vehicle theft, one of the primary crimes ALPRs should deter. Car theft surged nationally after 2019. Yet police cleared just 8.2 percent of reported vehicle thefts by arrest in 2023, the lowest rate of any offense category in the FBI data.
Our analysis produced three results. First, in the year following ALPR deployment, motor vehicle theft declined about 11 percent relative to comparison agencies. The decline held through the end of the year—a sign that it’s the cameras, not the publicity around them, driving the change. The same agencies’ other property crime did not fall at similar levels, which is what you would expect if the cameras were making the difference rather than some citywide shift in crime or reporting.
Second, arrests rose, with clearance of vehicle theft increasing about 16 percent. Complicating the story, the increase begins in the final months before the recorded activation date. That reflects the reality that departments often buy cameras as part of a broader push against vehicle theft. Still, the pre-trend suggests that not all of the increase in clearance can be attributed to the ALPRs themselves.
Third, while the cameras did not detectably accelerate the return of stolen vehicles overall, among vehicles that were recovered, the typical recovery arrived about six hours sooner, a small but significant change. Why didn’t the return rate for all stolen vehicles go up? The likely reason is that a camera-driven recovery requires a stolen car to pass an active camera with a readable plate that has been entered on a hot list, followed by an officer free to respond. Most recovered vehicles are instead located by their owners or found abandoned or crashed. And if they’re not found quickly, they’re often not found at all.
Our results match what research keeps finding about police sensor technology. As Robert VerBruggen noted in a City Journal piece on gunshot detection, sensors reliably deliver the narrow thing they produce (faster alerts, more evidence), but don’t always yield the broad outcomes cities desire. Similarly, CCTV reduces property crime in the places it watches and helps solve cases when someone actually retrieves the footage. Plate readers, on our evidence, raise the risk of driving a stolen car on covered roads and make it easier to establish where a vehicle went. That’s not a cure-all, but it’s not nothing, either.
City Journal contributing editor Rafael Mangual argued recently that the campaign to ban ALPRs relies on a misunderstanding of how they affect civil liberties, and that the proper remedy for these concerns is regulation, not prohibition. To judge the costs and benefits of such regulations, we need to know what the benefits of ALPRs really are. Until now, the research has been limited.
Our research does not, however, address the costs of ALPRs. We cannot say how often officers will turn the database on an ex, or what it costs a woman to learn her plate was run hundreds of times. It’s up to communities to weigh the real crime-reducing gains that we find against the privacy costs they may feel ALPRs impose.
As criminologists and former police officers, we know that police databases have always created opportunities for abuse. In 2016 the Associated Press counted more than 325 instances over three years of officers and employees fired, suspended, or forced out for misusing confidential law enforcement databases, many for running checks on exes and romantic prospects. Minnesota’s legislative auditor found in 2013 that more than half of the state’s 11,000 law enforcement database users had made queries suggestive of misuse; one former officer, whose driver’s license record was pulled more than 400 times by roughly 100 colleagues, collected over $1 million in settlements.
The new technology provided by firms like Flock does offer some tools for fighting abuse. Agencies can review their officers’ search logs, and the company now sells an automated assist that flags suspicious patterns for supervisors. That tool led one Georgia department to fire five officers this summer.
Still, abusers aren’t always accountable to their auditors. The Post identified a police chief as one of its culprits. In Milwaukee, one of the internal-affairs detectives assigned to investigate a colleague’s misuse of the department’s ALPR system was charged in July with felony misconduct for running twenty searches of his own, each logged as “test” or “training.” The man doing the auditing was himself the abuser.
The logs also depend on compliance from the very people whom they’re meant to monitor. In one Missouri city, a records request showed that 85 percent of searches were explained with just one word in the audit log, thus not containing enough information to be meaningful. Fewer than 3 percent were tied to a criminal case. Vendor audits also pose the possibility of a conflict of interest. A subscription-based business has an incentive not to embarrass the departments paying for its services.
What’s missing, in our view, is an external infrastructure for auditing ALPR networks. Such auditing is common in other parts of policing. Access to the FBI’s databases, for example, has for decades come with mandatory audits, run by the bureau and posing the threat of revoked access as punishment. Yet no equivalent regime exists for ALPRs.
Where such outside audits have happened, they often uncover abuses. In 2020, for example, California’s state auditor reported that none of the four agencies it examined had ever audited a single user search. A 2025 sample audit by the Illinois secretary of state caught Flock data being shared with ICE agents, a violation of Illinois state law. Making such outside audits mandatory and regular should be a major part of any ALPR reform agenda.
While an outside auditor would represent the most important change, communities weighing use of the technology should still demand limits to retention, tight rules on who searches and why, and data access for independent evaluation of effectiveness. Each variable interacts with the others. Short retention leaves less available to misuse, which lightens what an audit has to catch. Tight access shortens the list of users an auditor must watch, making a longer retention window possible. Balancing these priorities is the best way to prevent abuse while preserving investigative efficacy.
Any network that photographs a large number of public comings and goings should have to prove that it is maximizing public benefit and minimizing public cost. For the first time, we have a decent estimate of one of the benefits of ALPRs. That won’t settle the argument over the technology’s costs, but it should provide evidence to anchor those debates.