Faulty Facial Recognition Led to His Arrest—Now He’s Suing

Detroit police wrongfully arrested another Black man based on flawed facial recognition technology that often yields errors in identifying people of color, according to a new lawsuit obtained by Motherboard.

Michael Oliver, 26, was arrested for a larceny he didn’t commit after the software led the Detroit Police Department (DPD) to him in July 2019, according to the suit, filed Friday in Wayne County Circuit Court. Oliver's attorneys provided Motherboard with court documents for the case, which was first reported by the Detroit Free Press and is the second of its kind in Detroit reported this year.

“[Police relied] on failed facial recognition technology knowing the science of facial recognition has a substantial error rate among black and brown persons of ethnicity which would lead to the wrongful arrest and incarceration of persons in that ethnic demographic,” the lawsuit states. The case comes after the same police department was exposed by the New York Times  in June for wrongfully arresting 42-year-old Robert Williams after using the algorithmic system.

Oliver, whose story echoes Williams’ dramatic ordeal, is suing the city of Detroit and the white detective who allegedly botched the case for at least $12 million. He claims the algorithm along with “grossly negligent” police work led to his nearly three day imprisonment—and ultimately hijacked his life.

“It was wrong,” Oliver told Motherboard. “I lost my job and my car; my whole life had to be put on hold… That technology shouldn’t be used by police.”

Oliver—who has an oval-shaped face and a tattoo above his left eyebrow—only vaguely resembles the actual suspect, who has no visible tattoos, a rounder face and darker skin, according to photo evidence.

“The detective blew it, frankly,” said Oliver’s lawyer, David Robinson. “He certainly was in possession of a photograph of the real crook but for some reason wasn’t able to distinguish the difference between them.”

The case raises alarms about facial recognition technology—widely criticized as one of the most troubling forms of police surveillance—after police across the US used it to identify and arrest Black Lives Matter protesters earlier this summer. The New York Police Department, for example, used a facial recognition system to last month track down 28-year-old activist Derrick Ingram, who was accused of assault after allegedly shouting into a police officer’s ear with a bullhorn.

Oliver’s case of mistaken identity began on May 15, 2019 when a teacher driving by a group of brawling students used a cellphone to record the fight near a school in Detroit, according to Oliver’s arrest warrant.

A young black man involved in the scuffle spotted the teacher, Stephen Cassini, filming and reached into his car, grabbed his cellphone and cracked it, according to the warrant. Cassini captured a video of the incident on his phone before it was swiped and gave the footage to the DPD, according to the lawsuit.

Even though the department had “no written policy” allowing cops to use facial recognition technology, detective Donald Bussa asked a lieutenant to run an image of the suspect through a database of photos in search of a match, according to the lawsuit. The algorithm sent back a shot of Oliver and “Bussa assumed [he] was the person who snatched Cassini’s phone,” it states.

After running the software, developed by the firm DataWorks Plus, Bussa showed Cassini a photo lineup of suspects and the teacher allegedly identified Oliver. The detective then got a warrant for his arrest “with no further investigation,” the lawsuit states.

Oliver was driving to work on July 31, 2019 when he was pulled over, “handcuffed and carted off to jail,” according to the court documents. His car was impounded and he was forced to wait anxiously behind bars for 2 1/2 days—with little information about what he had supposedly done.

“I was nervous because I know people go to jail for things that they didn’t do,” Oliver said. “There was no explaining to police that it wasn’t me. They weren’t trying to understand that.”

In jail, he slept on a cement floor and ate bologna sandwiches—the only option—for breakfast, lunch and dinner, he said. While behind bars, “Bussa never made any attempt to take a statement from [Oliver] or do anything to offer [him] the chance to prove his innocence,” according to the lawsuit.

But there were soon signs that police had busted the wrong guy. Unlike the actual thief, Oliver had faded sleeve tattoos on both of his arms. That amount of ink, along with images of his face, should have been a dead giveaway, said Oliver’s criminal defense lawyer, Patrick Nyenhuis.

“He told me, ‘It wasn’t me.’ I looked at the pictures and said, “Yeah that’s obviously not you,” Nyenhuis told Motherboard.

Even so, during Oliver’s first court date in August 2019, law enforcement officials refused to back down.“The prosecutor wasn’t convinced. He said, ‘He could have gotten tattoos after the crime,’” Nyenhuis said. “I don’t know how they messed it up so badly.”

By mid-September, the Wayne County Prosecutor’s Office dropped the larceny charges, but Oliver says he’s still reeling. In the end, he suffered “embarrassment, indignation, outrage, shame, fear, anxiety, and loss of reputation” along with “economic loss,” according to the lawsuit.

Robinson now suspects the Detroit facial recognition technology has erroneously targeted many other people of color—and that more cases may soon emerge. “It’s not comprehensive to different cultures and colors,” he said of the software. “The designers of the algorithm are white folks, so it’s narrow; it works for the majority race.”

In the case of Robert Julian-Borchak Willams, the Wayne County Prosecutor’s Office apologized after the New York Times ran the story. Detroit Police Chief James Craig  later admitted the technology misidentified suspects in 96 percent of cases.

Oliver, for his part, won’t be satisfied with a mea culpa.

"He wants more than an apology. He wants justice—and he wants to right this wrong,” Robinson said.