Flock AI cameras misread a license plate on a Dodge Ram, mistaking the number seven for a two, and notified the police. The police then pulled over the driver, Brandon Upchurch, and released a dog on him, only to find out later that he was innocent. The incident, which occurred in Toledo, Ohio, highlights how AI cameras can potentially target non-criminals, causing unnecessary trauma and harm to them.
Recently reported by Business Insider, the story recounts an incident from 2024 in which Upchurch was driving back home with his cousin from a convenience store when police pulled him over on suspicion of driving a stolen vehicle.
Even as Upchurch was getting on the ground after receiving a warning, the officer, Adrian Wilson, released the dog on him. Then, “the dog latched onto Upchurch’s dreadlocks, rammed his head into the ground, and sunk its teeth into his arm. Wilson later said he thought Upchurch had tripped and was getting ready to run.”
One of the officers told Upchurch, “Your car has a stolen license plate on it.” Upchurch denied the allegation. He was taken to the hospital for the injuries before being put in the county jail for hours. Later, it came to be known that Upchurch was telling the truth- his truck wasn’t stolen.
The Flock AI camera, built by technology startup Flock Safety, misread the license plate on Upchurch’s Dodge Ram and pinged the police about the allegedly stolen truck. An error in reading the license plate led the police to make a wrongful arrest.
Eventually, the two charges Upchurch faced, of obstructing official business and resisting arrest, were dismissed. However, the question is, who is held accountable for the wrongful arrest of an innocent individual?
Not a First Time for the Police
Unfortunately, the incident involving Upchurch isn’t an isolated one. The report further states that “in a dozen instances, misreads by Flock’s automated license plate readers, or a lack of verification by officers, resulted in people who hadn’t committed crimes being stopped at gunpoint, sent to jail, or mauled by a police dog, among other outcomes.”
Flock refused to reveal misread rates but clarified that cases that are marked as misread by customers are used to improve its system. A company spokesperson said in a statement:
“We have strong confidence in our models and continuously improve them through ongoing optimization and testing.
“We respect and value concerns and feedback raised by community members and are continuously building on our technology.
“We have the utmost sympathy for any victim of harm — and that is why we are committed to the work we do to solve and deter serious crimes and provide security for communities.”
