In January 2020, Robert Williams spent 30 hours in a Detroit jail because facial recognition technology suggested he was a criminal. The match was wrong, and Mr. Williams sued.
On Friday, as part of a legal settlement over his wrongful arrest, Mr. Williams won a commitment from the Detroit Police Department to do better. The city adopted new rules for police use of facial recognition technology that the American Civil Liberties Union, which represented Mr. Williams, said should become the new national standard.
“Hopefully it moves the needle in the right direction,” Williams said.
Mr. Williams was the first known person to be wrongfully arrested because of facial recognition errors. But he wasn’t the last. Detroit police have arrested at least two other people following botched facial recognition searches, including a woman accused of carjacking when she was eight months pregnant.
Law enforcement agencies across the country are using facial recognition technology to try to identify criminals whose misdeeds are caught on camera. In Michigan, the software matches an unknown face to those in a database of mug shots or driver’s license photos. In other jurisdictions, police use tools like Clearview AI that search photos pulled from social media sites and the public Internet.
One of the most notable new rules adopted in Detroit is that images of people identified through facial recognition technology can no longer be shown to an eyewitness in a mug shot unless there is other evidence linking them to the crime.
“The 'take a photo, line it up' system will end,” said Phil Mayor, an attorney for the ACLU of Michigan. “This settlement moves the Detroit Police Department from being the most documented misuser of facial recognition technology to a national leader in having safeguards in its use.”
Police say facial recognition technology is a powerful tool to help solve crimes, but some cities and states, including San Francisco; Austin, Texas; and Portland, Oregon, have temporarily banned its use due to concerns about privacy and racial bias. Stephen Lamoreaux, chief of information technology at Detroit’s criminal intelligence unit, said the police department is “very keen to use the technology in a meaningful way for public safety.” Detroit, he said, has “the strongest policy in the nation right now.”
How it goes wrong
Mr. Williams was arrested after a crime in 2018. A man stole five watches from a boutique in downtown Detroit while being caught on surveillance camera. A loss prevention company provided the footage to the Detroit Police Department.
A search of the man’s face against driver’s license photos and mug shots returned 243 photos, classified based on the system’s confidence that they were the same person in the surveillance video, according to documents released as part of Mr. Williams’ lawsuit. An old driver’s license photo of Mr. Williams was ninth on the list. The person who ran the search deemed it the best match and sent a report to a Detroit police detective.
The detective included Mr. Williams' photo in a “six-pack photo lineup” (photos of six people in a grid) that he showed to the security contractor who had provided the store's surveillance video. He agreed that Mr Williams was closest to the man in the boutique, and this led to the arrest warrant. Mr Williams, who was at his desk at an automotive supply company when the watches were stolen, spent the night in jail and had his fingerprints and DNA taken. He was charged with retail fraud and had to hire a lawyer to defend himself. Prosecutors ultimately dropped the case.
He sued Detroit in 2021 hoping to enforce a ban on the technology so others wouldn't suffer its fate. He said he was shocked last year when he learned that Detroit police had charged Porcha Woodruff with carjacking and robbery after poor facial recognition. The police arrested Ms. Woodruff while she was preparing her children for school. He also sued the city; the case is ongoing.
“It’s so dangerous,” Williams said, referring to facial recognition technology. “I don’t see any positive benefit from it.”
The new rules
Detroit police are responsible for three of seven known cases in which facial recognition led to an unjust arrest. (The others occurred in Louisiana, New Jersey, Maryland and Texas.) But Detroit officials said the new controls would prevent further abuse. And they remain optimistic about the crime-solving potential of the technology, which they now use only in serious crime cases, including assault, murder and home invasions.
James White, Detroit's police chief, blamed the wrongful arrests on “human error.” His officials, he said, relied too much on leads produced by technology. It was their judgment that was faulty, not the machine's.
The new policy, which goes into effect this month, should help. Under the new rules, police can no longer show a person’s face to an eyewitness based solely on facial recognition.
“There has to be some sort of corroborating secondary evidence that is unrelated before there is sufficient justification to go to training,” he said. Mr. Lamoreaux of the Detroit Criminal Intelligence Unit. Police would need location information from a person's phone, for example, or DNA evidence, something more than a physical resemblance.
The department is also changing the way it conducts photo screenings. It’s adopting what’s called double-blind sequential screening, which is considered a fairer way to identify someone. Instead of presenting a “six-pack” to a witness, an officer — one who doesn’t know who the primary suspect is — presents the photos one at a time. And the lineup includes a photo of the person different from the one that came up with the facial recognition system.
Police will also have to disclose that a facial search was conducted, as well as the quality of the image of the face searched (how grainy was the surveillance camera? How visible is the suspect's face?) because a poor-quality image is less likely to produce reliable results. They will also have to disclose the age of the photo that came up through the automated system and whether there were other photos of the person in the database that did not match.
Franklin Hayes, Detroit's deputy police chief, said he was confident the new practices would prevent future misidentifications.
“There are still some things that could slip through, for example, identical twins,” Mr. Hayes said. “We can never say never, but we think this is our best policy yet.”
Arun Ross, a computer science professor at Michigan State University and an expert on facial recognition technology, said Detroit’s policy was a good starting point and that other agencies should follow suit.
“We don't want to trample on the rights and privacy of individuals, but we also don't want crime to become rampant,” Ross said.
How much does it help?
Eyewitness identification is a daunting task, and police have embraced cameras and facial recognition as tools more reliable than imperfect human memory.
Last year, Chief White told local lawmakers that facial recognition technology had helped “get 16 murderers off the streets.” When asked for more information, police department officials did not provide details about those cases.
Instead, to demonstrate the department's successes with this technology, police officers played surveillance video of a man spraying fuel inside a gas station and setting it on fire. They said he had been identified with facial recognition technology and arrested that night. He later pleaded guilty.
The Detroit Police Department is one of the few that monitors facial recognition searches, submitting weekly reports on its use to an oversight committee. In recent years, it has averaged more than 100 searches a year, about half of which have yielded potential matches.
The department tracks only how often it gets a lead, not whether the lead is completed. But as part of the deal with Mr. Williams — who also received $300,000, according to a police spokeswoman — it must conduct an audit of his facial recognition searches dating back to when he began using the technology in 2017. If it identifies others In cases where people have been arrested with little or no corroborating evidence other than a facial match, the department is supposed to alert the appropriate prosecutor.
Molly Kleinman, director of a technology research center at the University of Michigan, said the new protections look promising, but she remains skeptical.
“Detroit is an extraordinarily policed city. There are cameras everywhere,” she said. “If all this surveillance technology actually did what it says it would, Detroit would be one of the safest cities in the country.”
Willie Burton, a member of the Board of Police Commissioners, an oversight group that approved the new policies, described them as “a step in the right direction,” though he still opposed the use of facial recognition technology by police.
“The technology is not there yet,” Burton said. “One false arrest is one too many, and having three in Detroit should be a wake-up call to stop it.”