It’s no secret that law enforcement’s use of facial recognition technology disproportionately affects Black people more often than it does any demographic group…and it’s been like that for years. But one city plans on making changes to how it uses the technology, thanks to the wrongful arrest of a Black man nearly six years ago.

Last month, along with agreeing to pay Robert Williams $300,000 in a settlement agreement, the City of Detroit agreed to revise how police use facial recognition to solve criminal cases, according to the Associated Press.

The Detroit police will now go back and look at all cases between 2017 and 2023 in which facial recognition was used and will notify a prosecutor if an arrest was made without evidence independent of the technology.

Robert Williams’ wrongful arrest

Williams was arrested in 2018 after someone stole watches from a store in the Detroit area. The only evidence Detroit Police had was an image from the store’s surveillance footage. Officers then sent the image to the Michigan State Police so they could run a search using face recognition technology, and what came up was an expired driver’s license photo of Williams, according to the lawsuit.

Although Williams said he was not the man in the image, Detroit police still used his photo to create a photo lineup and show it to a man who was not a witness to the crime and only saw the store’s surveillance footage.

Despite this, an officer, who has not been identified, still applied for an arrest warrant.

In April 2021, Williams filed a civil rights lawsuit against the detective, the city and the city’s chief of police with the help of the ACLU and the University of Michigan’s Law School’s Civil Rights Litigation Initiative.

More from the ACLU:

The lawsuit alleges that the detective, through his omissions in the warrant application, misled the magistrate judge, resulting in issuance of an arrest warrant without the required probable cause. It also alleges what discovery in Mr. Williams’ case and several more recent facial-recognition false arrests in Detroit have since made obvious: That the city lacked any policy for law enforcement use of face recognition technology at the time the technology was used in this case, and that Detroit failed to train its police officers on the dangers of misusing face recognition technology in their investigations.

Past examples of facial recognition affecting Black people

Even before Williams’ wrongful arrest, facial recognition technology had already been doing Black people dirty.

In 2018, 28 members of Congress, including six members of the Congressional Black Caucus, were falsely identified as suspects charged with a crime on Amazon’s facial recognition technology.

In December 2020, a Black man in New Jersey filed a lawsuit after he was wrongly identified as a suspect who shoplifted from a hotel gift shop. He spent 10 days in jail.

In July 2021, a Black girl was banned from a Michigan skating rink after facial recognition software misidentified her for a different Black girl who got in a brawl at the business on a prior date.

In August 2023, a pregnant Black woman was falsely identified and arrested as a carjacking thief in Detroit thanks to the technology.