City Sued Over Multiple Erroneous Flock LPR Camera-Based Stops

Three individuals — two whose license plate was allegedly misread by a Flock Safety LPR camera and a third whose vehicle was allegedly mistaken for another by an officer reviewing Flock camera footage – are suing a New Mexico city over traffic stops.

The three people, two of whom were minors, were treated as potentially violent suspects and held at gunpoint before police recognized the errors and released them, according to complaints filed in a New Mexico state court.

In the report, we examine the lawsuits and the incidents at issue, LPR error rates, the responsibility of police, and Flock's response to the allegations.

Executive Summary

As police departments increasingly rely on LPR technologies, the systems introduce new risks of wrongful stops and arrests based on misread license plates. Accuracy rates vary depending on the system, the type of license plate lettering, the design of the plate, and whether license plates are obscured.

However, the police should be able to verify LPR results relatively easily since they are short strings of alphanumeric characters that can be read by a police officer in their vehicles to ensure they are not wrongfully arresting people based on an LPR system error. This is fundamentally different and more reasonable than, e.g., facial recognition, where it is far harder to determine if a person matches a facial recognition system alert.

While Flock is not a defendant in the Espanola lawsuits, it faces legal claims in another case over an alleged wrongful stop based on an LPR-related error. An Ohio man, Michael Smith, alleged he was stopped and detained for 40 minutes based on mistaken information transmitted via Flock to officials asserting that he was a “suspect” on a “hot list.” The officers released Smith after learning he was not wanted in connection with an investigation.

Even when accuracy is high, errors by LPR systems and in using LPR systems represent a significant liability risk for police departments and a threat to civil liberties. Every single LPR error resulting in a wrongful detention, arrest, or use of force by police has the potential to become a legal claim against municipalities. As the Ohio case shows, LPR companies may also be sued over the errors.

Representatives for the City of Espanola did not reply to our requests for comment.

Background

• Flock Warns Custom Hot Lists Are Not Probable Cause
• Flock Sued By Man Detained Over LPR "Hot List" Error
• Flock Responds to "Hot List" Error Lawsuit
• The Power of National LPR Watchlists (Flock and Motorola) Examined
• Flock Raises Another $150 Million, Valuation Now At $3.5 Billion

Alleged Flock Camera Error

One of the stops in the recent lawsuits against Espanola, New Mexico, involved two sisters, one of whom was 12 years old, whose license plate was allegedly misread by a Flock camera, according to their legal complaints (See the younger sister's complaint here and the older sister's complaint here).

The older sister was driving her Kia Optima, and the younger sister was riding in the passenger seat on July 23, 2022, when they passed through an intersection where a Flock camera was mounted, according to the complaints. The sisters alleged that the camera misread their license plate, BLGP02, as BLGP07 – a license plate for a vehicle that was reported as stolen.

Alerted to a “match” for a stolen vehicle, police officers stopped the sisters' car. A police sergeant read the correct plate number over the phone to a dispatcher when pulling the sisters over, but he did not wait for confirmation of whether it matched the Flock alert before proceeding with the traffic stop, according to the complaints.

With a gun drawn, an officer ordered the older sister out of the car and ordered her to kneel on the pavement while another officer ordered the younger sister out, handcuffed her, and put her in the back of a police cruiser, according to the complaint.

Police allegedly contacted dispatchers a second time, providing the incorrect plate number from the Flock alert instead of the correct plate from the physical license plate, before eventually providing dispatchers with the vehicle's VIN, according to the complaints.

The VIN allegedly revealed the true license plate number of the car and that a mistake had been made by the Flock equipment, according to the sisters’ complaints.

The complaints continued:

The officers then hypothesize that the FLOCK camera must have 'read' the Plaintiff's license plate incorrectly because a clear license plate cover obscured, in their opinion, the bottom of the number 2, making it appear to be a 7.

Eventually, Sergeant Martinez determined that Plaintiff's vehicle did not bear the stolen license plate that Rio Rancho Police had reported.

Sergeant Martinez uncuffed the Plaintiff's sister, explained that he had been mistaken, then told the Plaintiff that the clear cover on her license plate made the last number look like a '7' and not the '2' that it was. [Emphasis Added.]

Alleged Officer Error

Another lawsuit against Espanola stems from a separate traffic stop involving a 17-year-old high school student. The teenager was driving a white Toyota Tacoma truck at 11:00 am on Aug. 15, 2022, when police received a “BOLO” (“be on the lookout”) alert about his vehicle, claiming that it had been implicated in a series of armed robberies, according to his complaint.

An officer conducted a “felony stop” of the teenager, ordering him at gunpoint to get out of the car, kneel on the pavement, and place his hands on top of his head, his complaint alleges. The teenager was handcuffed, “frisked from head to toe,” and detained until the officers determined that he was not the individual wanted in connection with the robberies, according to the complaint. At that point, he was released.

A lawyer for the teenager as well as the two sisters, Sheri Raphaelson, told IPVM that an officer issued the "BOLO" after viewing footage from a Flock LPR camera and mistaking the vehicle for the wanted vehicle. The stop was not based on a license plate match. She provided IPVM with a photo of the footage.

Raphaelson told us:

The officer identified [the teen’s]'s truck by reviewing FLOCK footage and seeing a truck that he wrongly believed matched the description of the suspect's vehicle. There were differences in the description of the suspect's vehicle and [the teen’s]'s vehicle (as seen in the FLOCK footage) but the officer wrongly concluded that they were the same truck. Human error by the officer using the FLOCK system to review vehicles that had passed through an intersection near the crime scene. The officer then used the photo FLOCK snapped of [the teen’s]'s truck to create a BOLO for [the teen’s]'s truck. [Emphasis Added.]

Tort Claims Act

All three individuals are suing the city under the state’s Tort Claims Act, which establishes circumstances when people can recover damages from local government employees for personal injury. The complaints alleged that the police officers’ negligent acts resulted in violations of Constitutional rights against illegal search and seizure, caused harms including “anxiety, fear, worry, and restriction of movement.”

A section about law enforcement liability, 41-4-12, states:

The immunity granted pursuant to Subsection A of Section 41-4-4 NMSA 1978 does not apply to liability for personal injury, bodily injury, wrongful death or property damage resulting from assault, battery, false imprisonment, false arrest, malicious prosecution, abuse of process, libel, slander, defamation of character, violation of property rights or deprivation of any rights, privileges or immunities secured by the constitution and laws of the United States or New Mexico when caused by law enforcement officers while acting within the scope of their duties. [Emphasis Added.]

Reasonable Suspicion

To comply with the Fourth Amendment of the Constitution, police must have "reasonable suspicion" to make a traffic stop or detain someone.

As interpreted by courts, "reasonable suspicion" means that police officers must observe specific indications that an individual has either committed a crime or is in the process of committing one. Examples include an officer witnessing traffic violations such as speeding or noticing an individual driving recklessly.

If officers receive information showing that a vehicle matches a detailed description of one that has been linked to a crime, the information has also been found to constitute reasonable suspicion. However, police department policies typically require officers to confirm LPR alerts before making a stop.

A 2017 LPR policy template created by the Department of Justice's Bureau of Justice Assistance states the following:

Receipt of an LPR alert for a stolen or felony vehicle may not rise to the level of reasonable suspicion and is not sufficient probable cause to arrest without confirmation that the alert is still valid and active. If the alert is for another type of transaction, the officer will read the description of the alert and follow the appropriate action or reporting method. If an LPR alert cannot be verified both visually and for validity, then law enforcement should not act on the alert and it should be rejected. [Emphasis Added.]

Ruling On ALPR

The DOJ Bureau of Justice Assistance policy followed a precedential 2014 ruling by the Ninth Circuit Court of Appeals, which reinstated a case against San Francisco police that had been thrown out by a lower court. The appeals court suggested that ALPR “hits" needed to be verified to constitute reasonable suspicion.

The case was on behalf of a 47-year-old woman with no criminal record who was held at gunpoint by police and handcuffed after an ALPR reader falsely identified her vehicle as stolen. Officers failed to visually confirm whether her plate matched the stolen vehicle before initiating a stop, according to the appeals court opinion. The lawsuit was sent back to a lower court and dismissed by mutual agreement of the parties in 2015.

A higher standard than reasonable suspicion, probable cause, is required under the Fourth Amendment to make an arrest. Flock informed customers in an August 2023 user guide disclaimer that its “custom hot list” alerts, “alone” should not be used as probable cause. It did not include any warnings about whether its alerts can amount to reasonable suspicion.

Flock Response

While it is not a defendant in the lawsuits, Flock was implicated in both cases, and an alleged malfunction of its technology was blamed for one of the stops. We contacted Flock and requested comment on the allegations and information on its accuracy rates. A representative for the company replied:

Flock Safety technology is best-in-class, consistently performing above other vendors in side-by-side tests on plate reads and accuracy. In addition, Flock encourages agencies to adopt a policy to double-check all alerts, both against the NCIC or other hotlist and via manual, human verification of the plate/vehicle in question.Given that neither of these lawsuits name Flock as a party, we have no comment on either. [Emphasis Added.]

When asked for specifics about Flock's performance in the referenced tests, the spokesperson replied:

Unfortunately, it isn't that simple. The accuracy of any public safety technology may vary depending on a myriad of factors including location, time of day, weather, plate design, volume of traffic, angle of the devices, and what a user is trying to identify (plate number, color, make, model, etc). Flock Falcon LPRs consistently perform at best-in-class levels and our team is constantly improving our machine learning algorithms. [Emphasis Added.]

LPR Error Rates

With police departments gathering billions of license plate scans in the US, the impact of even small error rates quickly escalates. The Electronic Frontier Foundation found in a survey that 173 law enforcement agencies collectively scanned about 2.5 billion LPR images in 2016 and 2017 combined. We project current rates are many multiple times higher as LPR usage has exploded in the last five years, including with the astronomical growth of Flock.

IPVM recently conducted the world's first and only LPR comparative test - LPR Rankings - Avigilon Alta, Axis, Genetec, Hanwha, Motorola, Rekor, Verkada (though Flock refused to allow us to buy any Flock equipment for this test).

Our research shows that while systems may have accuracy rates near 100% in optimal conditions, accuracy for all systems dropped 10 to 20 percentage points when plates were partially covered or otherwise obscured.

Conclusion

While LPR cameras provide law enforcement with a rapid means of identifying vehicles tied to crime, offering public safety benefits, over-reliance on the systems introduces the potential for errors leading to wrongful stops and arrests.

Because LPR "hot lists" are used to track vehicles that are either stolen or involved in major offenses such as kidnapping and armed robbery, police are more likely to employ "felony" stop tactics for those vehicles — positioning themselves with guns drawn and treating a suspect as though they are armed and dangerous. With adrenaline running high, felony stops can result in officers shooting and even killing unarmed suspects over misunderstandings or non-compliance.

The lawsuits against Espanola show how police rushing to act on erroneous LPR "hits" or camera footage may not follow protocols to verify the information. They demonstrate the importance of high accuracy from LPR providers and more rigorous checks by the police to prevent wrongful stops.