The Research Is In.
ALPR Doesn't Work.

The Cambridge Journal of Evidence-Based Policing published a comprehensive review of ALPR research in December 2025. Its conclusion: "the jury is still out on whether ALPRs are more effective in achieving public safety goals." The review noted that most existing studies measured efficiency (how fast officers can check a plate) rather than crime-control effectiveness (whether crime actually goes down).

This is the single most authoritative assessment of the ALPR evidence base. It was written by criminologists, published in a peer-reviewed journal focused on policing, and it could not find sufficient evidence that the technology works.

Three methodologically rigorous evaluations of ALPR programs have been conducted: Mesa, Arizona (Taylor 2012), Alexandria and Fairfax County, Virginia (Lum 2011), and the LSU entertainment district in Baton Rouge (Kernahan & Valasik 2019). All three used mobile or fixed ALPRs. All three found no significant crime reduction.

These are not opinion pieces. They are controlled studies published in criminology journals. The technology has been tested, and it has not passed the test.

In 2025, researchers studied Atlantic City, New Jersey, where a police department installed fixed ALPR cameras on every entrance and exit to a barrier island. This is about as close to total coverage as you can get. Every vehicle entering or leaving the city was captured.

The result: no reduction in violent crime.

If blanket coverage on a geographically contained island doesn't reduce crime, more cameras in Apex won't either. The theory that coverage equals deterrence does not survive contact with the data.

Multiple studies (King 2008, Cerezo 2013, Gómez 2021) have found that localized camera networks relocate crime to areas without coverage rather than reducing it overall. The crime doesn't disappear. It shifts to wherever the cameras aren't.

For Apex, this means cameras on a few corridors may push activity to residential streets, neighboring jurisdictions, or anywhere else outside the camera's view. The net effect on total crime is zero or close to it.

As a surveillance network expands, new cameras get placed in progressively less critical areas. Each additional camera deters less than the one before it. This directly undercuts the "more coverage equals more safety" argument that drives expansion contracts.

Apex currently has 7 cameras. The unsigned contract adds 10 more. The research says those 10 will accomplish less than the original 7 already didn't accomplish.

Note the language: "contributing to," not "solving." A camera that captured a plate near a crime scene counts as a contribution, even if detectives would have solved it anyway. At 0.05%, ALPR is a rounding error in the crime-solving picture, and this is in a city with one of the largest camera networks in the country.

The broader CCTV evidence base finds that cameras are most effective only when combined with improved lighting, security guards, and environmental design changes. The camera by itself isn't doing the work. It's the full package of interventions, and the camera gets the credit.

When APD claims ALPR is "already reducing crime," they are attributing an outcome to a single technology without controlling for anything else that changed.

The Electronic Frontier Foundation's Street Level Surveillance project maintains the most comprehensive vendor-agnostic assessment of ALPR technology. It covers the entire field: Motorola/Vigilant, Flock, Rekor, Elsag, Axon, Perceptics, and Jenoptik.

The EFF's assessment: aggregate ALPR data "can paint an intimate portrait of a driver's life and even chill First Amendment protected activity." That isn't a statement about Flock's data-sharing practices. It's a statement about what the technology itself makes possible regardless of which company's name is on the hardware.

Perceptics was the ALPR vendor for U.S. Customs and Border Protection, one of the best-funded law enforcement agencies in the country. It was hacked, and its data was published online.

The promise that any vendor will keep this data secure doesn't hold up against the track record. If CBP's vendor couldn't protect the data, the "we'll secure it" assurance from any ALPR company should be evaluated with that context in mind.

A randomized vignette study found that public support for police surveillance technology depends on a clear purpose and a "discernible victim." When the objective is vague or there is no specific victim, public apprehension about surveillance motives increases sharply.

ALPR's blanket-everyone model, where every vehicle is captured whether or not it is connected to a crime, is exactly the configuration that generates the lowest public support.

A 2025 study of California police departments found that surveillance technology adoption is predicted by police-community racial asymmetry, with "much weaker support for crime control theories." In plain language: the data suggest these systems are adopted based on the demographics of a community, not its crime rate.

That finding matters because it explains a pattern: crime data never justifies these programs after the fact, because crime data wasn't the reason they were purchased in the first place.