A blue wooden sign with golden engraved letters reads "City of Syracuse, Dept. of Police" on a green lawn outside of the police station downtown. A blue sky shines down on the scene.
Syracuse Police headquarters on South State Street in downtown Syracuse. Credit: Julie McMahon | Central Current

When Syracuse’s Common Council greenlit the city’s use of Flock license plate readers, a majority of councilors did not know of research that found the readers regularly spit out incorrect information.

According to research published three years ago by a security and surveillance research firm, Flock license plate readers misread the state of 1-in-10 license plates. That does not include any other reading errors that may have happened. 

“It’s something that definitely needs to be brought up,” said councilor Rita Paniagua. “I had no idea that there’s a percent of misreads.”

At least four councilors expressed concern over the error rate from the study in interviews with Central Current. The council gave Flock permission to put up 26 license plate readers throughout the city on July 1. 

IPVM, the firm that released the study about Flock readers, also found the readers logged plate reads multiple times, vehicle characteristics could be misread and that the car make was regularly incorrect. Flock stopped selling its readers to IPVM for testing shortly after the report was issued. Flock is the only license plate reading technology provider who has refused to allow IPVM to test its product, said Donald Maye, IPVM’s head of operations.

In a statement to Central Current, Flock did not dispute the results of IPVM’s study, instead alleging that reviewers do not test the readers in enough settings, according to a spokesperson. Earlier this year, Flock retracted a statement about the IPVM research after IPVM refuted Flock’s claims. 

Flock defended its product as helping increase the rate at which departments solve crime, citing a study done on its technology and police department clearance rates.

The misreads underscore long-held concerns among security experts and community members. Experts also worry about the potential violation of city residents’ privacy and constitutional rights. 

In other cities, misreads and misinterpretations of license plate data have had dire consequences for innocent people. At times, drivers have been held at gunpoint, attacked by K9s or falsely arrested. 

Syracuse police Deputy Chief Richard Shoff believes the department needs the readers to solve crime and solve it more quickly. The department believes readers could help in stolen vehicle and murder cases.

Police Chief Joe Cecile has touted an existing LPR at the corner of Salina and Colvin Streets as helping to solve multiple homicides. 

The two councilors who co-sponsored the July 1 legislation, Chol Majok and Amir Gethers, dismissed the error rate as typical of police technology. 

“10% is a fairly low number,” Gethers said. 

Experts worry about data retention and sharing

Technology experts, privacy advocates and community members believe the department’s policies for the retention of license plate reader data are inadequate and that use of the readers could infringe on residents’ rights. 

Their primary concerns are the city’s 30-day data retention period and that Flock has openly touted its data sharing.

Three members of the city’s Surveillance Technology Working group, which evaluates government technology that could result in the surveillance of residents, were interviewed for this story. All believe the retention period should at least be cut in half. 

Longer storage periods open the data up to potential hacks and misuse, said Johannes Himmelreich, a professor at Syracuse University whose research centers on ethics in technology. He sat on the working group. 

“Even if SPD is doing everything right, and I expect them to do everything right, the vendor might have their own policies or data leaks that might be problematic,” Himmelreich said. 

Some states have even shorter retention periods. In Vermont, readers are only allowed to retain data for three minutes.  

Daniel Schwarz, a privacy and technology strategist with the New York Civil Liberties Union, worries about nefarious uses. He believes the readers could be used to surveil protesters, undocumented citizens, abortion seekers from states where abortions have been banned and people of various faiths. 

Flock has drawn attention for its willingness to allow data sharing across jurisdictions, including with private organizations like homeowner associations, said working group member and data consultant Mark King. 

King wrote a dissenting opinion to the group’s approval of the technology. He opined in the opinion about whether the network of cameras Flock has built is constitutional. 

Shoff downplayed the department’s ability to share data and potentially help out-of-state police departments crack down on abortion seekers. 

If officers misuse the technology, they could be subject to an internal investigation, Shoff said. 

Data can only be kept longer than 30 days when it’s been flagged as evidence in an active investigation, Shoff said. He also said that a mayoral executive order prevents the department from sharing such data with U.S. Immigration and Customs Enforcement. 

“Just to be frank, some of our detectives would like to keep the data a lot longer,” Shoff said. 

Flock openly advertises data sharing between departments as an advantage to using their readers. Flock’s plate readers were able to scan more than one billion license plates each month as of 2021, according to a former version of the company’s website. 

The American Civil Liberties Union detailed in a 2022 research paper how Flock’s business model hinges on expanding its surveillance network. Maye, the IPVM head of operations, believes Flock has taken off in recent years because of its ability to share license plate data. Part of Flock’s appeal is that the more departments and private organizations use Flock’s technology, the more information is available to the departments, he said.

Richard Levy, a former member of Syracuse’s Citizen Review Board, who evaluated allegations of misconduct against the department, is dubious of the department’s ability to keep the data safe and worries whether the department can keep officers to use the rules it set. 

In Wichita, Kansas, a police officer used Flock cameras to stalk his ex-wife, according to reporting from the Wichita Eagle and TV station KAKE.  

“I don’t think I’ve read about any oversight on the collection of this information that would make me feel comfortable with this being put into practice,” Levy said. 

How the cameras work

Each Flock Safety LPR camera is a stationary reader that scans the license plate of every vehicle that passes by. 

The readers have a field of vision of 20 feet, and can read license plates up to 75 feet away. 

Once scanned, each license plate number is compared against a “hotlist” of plate numbers connected to or suspected to be connected to a crime, according to Flock’s website. The company states on its website that the hotlist is a database from the FBI’s National Crime Information Center. If a license plate matches a plate number on the hotlist, law enforcement is notified. 

Each scanned license plate, a picture of the car and details about the car are uploaded to a database and stored for 30 days, whether the plate is a hit or not. Every Syracuse police officer will have a log-in to the Flock system, Shoff said. However, only detectives will be using the database, Shoff said. Flock keeps data indefinitely on which officers have accessed the system. 

While the department has publicly said the LPR data will be used exclusively for violent crimes and car thefts, Shoff frequently veered into discussion of other uses for the data in an interview with Central Current. He said the technology could even be useful for missing persons cases and specific traffic violations. 

“It’s a little kind of creepy, but kind of the way the world right now,” Shoff said of license plate readers.

LPR and database errors have dire consequences

License plate readers misreading license plates and errors in hot list databases have led to drivers being held at gunpoint, attacked by police dogs or falsely arrested, media reports from across the country show. 

Central Current was able to find at least five recent incidents – four involving Black drivers – of volatile traffic stops stemming from LPR misreads or errant hot list data.

In 2021, officers in Charlotte, North Carolina searching for a stabbing suspect entered the wrong name in their license plate reader system, leading them to hold an innocent Black woman at gunpoint and falsely arrest her, WCNC TV reported. 

In 2022, police in Frisco, Texas held a Black family at gunpoint after a license plate reader correctly matched a car’s plate to a plate entered into a database of stolen cars, the Washington Post reported. An officer had incorrectly entered the stolen car’s plate into the system, according to the Washington Post. 

On New Year’s Day 2024 in Tampa Bay, a father and his daughter, who has epilepsy, were handcuffed and held at gunpoint after an officer made a typo while running a license plate through an LPR database, according to WTSP TV in Tampa. 

Other traffic stops based on incorrect LPR data have escalated even further. In April of this year, Toledo Police Department officers using a Flock reader pulled over a vehicle based on an errant read from the LPR, according to The Toledo Blade. The vehicle’s owner, a Black man, was attacked by a police dog and accused of resisting arrest and obstructing official business. 

Myrtle Beach police held a Black teenager at gunpoint and handcuffed her in May after a license plate reader led officers to stop her vehicle. The teenager has since filed a civil rights lawsuit alleging the police wrongfully detained her – and that officers knew before drawing their weapons that they had pulled over the wrong vehicle.

Flock touts a “vehicle fingerprint” concept on its website that it believes would guard against misidentifications by providing a plate number, car make, color and decals. 

King, the data analyst, said the technology working group put forward stipulations to help ensure the police department confirms the accuracy of a reading before situations escalate.

He believes any error rate is still alarming. 

“Such a high error rate still worries me.  What if the procedures aren’t always followed properly?” King said.

“Generally speaking I am very comfortable with this FLOCK LPR technology not being 100% accurate,” Shoff wrote in an email. “As you know there are not any technologies, at least that I know of, that are 100% accurate.”

read more of central current’s coverage

Patrick McCarthy is a staff reporter at Central Current covering government and politics. A graduate of Syracuse University’s Maxwell and Newhouse Schools, McCarthy was born and raised in Syracuse and...