False facial recognition sends innocent New Jersey man to jail

(CNN) - New Jersey police arrested an innocent man on unlawful possession of a weapon, resisting arrest and aggravated assault charges after he was falsely identified by facial recognition technology.

Law enforcement agencies across the country use facial recognition to generate leads on suspects, but it can have a major negative impact on innocent people if the technology gets it wrong.

Police responded to reports of a theft at a Hampton Inn in Woodbridge, New Jersey.

The suspect drove away, hitting a police car and, police claim, almost running over an officer.

Using cutting-edge facial recognition technology, police quickly arrested Nijeer Parks, a 31-year-old Paterson resident.

At the time the crimes were being committed at the Hampton Inn in January 2019, Parks was 30 miles away in a Western Union, sending money to his fiancée.

Police found what they say was a fake driver’s license at the scene. Using the photo on the ID, they got a facial recognition match to Nijeer Parks.

Based on that evidence, a judge signed a warrant for Parks’ arrest.

“The thing that is so notable about this case isn’t that the facial recognition got it wrong. It’s that we found out,” Albert Fox Cahn, Executive Director of Surveillance Technology Oversight Project, said.

Parks has now become the face of an issue researchers and privacy advocates have been warning about for years: facial recognition in the hands of law enforcement.

“When we talk about the number of facial recognition scans unfolding in the United States every day, we don’t even know the full number,” Cahn said. “You can have people who are being sent to jail wrongly who never know that facial recognition played a role in their arrest.”

A 2019 federal report found widespread evidence of racial bias in nearly 200 facial recognition algorithms that were far more likely to misidentify people of color than they were white people.

“If you keep showing a model a very specific type of face, it will distinguish different faces using very specific features so it ends up maybe focusing on attributes of a face that don’t really distinguish faces of a different culture or of a different skin type in the case of some of the products that we tested which struggled with performing well on darker skin types,” researcher Deb Raji said.

Parks spent 10 nights in jail. His case wasn’t dropped until almost a year later.

He’s now suing those involved in his arrest.

“Facial recognition is one of those products, where it is dangerous when it doesn’t work and it is dangerous when it works. It is still weaponized against people of color. It’s a technology that is disproportionately used on that group,” Raji said.

Facial recognition is being used throughout the country.

In New York City alone, police said they identified roughly 2,500 possible facial recognition matches in 2019.

“In many jurisdictions, facial recognition scans target the databases of past mug shots. This is really crucial because when you’re only looking at mug shots from those people who were arrested in the past, you’re building facial recognition on infrastructure that comes from decades of biased policing,” Cahn said.

Advocates say it helps the police solve crimes by quickly generating leads from nothing more than a photograph or video, but critics say those leads are often used as evidence.

“When police use this technology and they get it wrong, the worst-case scenario isn’t a night in jail. The worst-case scenario isn’t being wrongly convicted,” Cahn said. “In the aftermath of last year’s police killings and so many others before, we know at this moment that when police make mistakes Black Americans die.”

The debate over law enforcement’s use of facial recognition technology comes amid the ongoing debate over police use of force.