In January 2020, Robert Williams spent 30 hours in a Detroit jail after facial recognition technology identified him as a criminal. The match was incorrect, and Williams sued.
On Friday, Williams was promised improvements by the Detroit Police Department as part of a court settlement over his wrongful arrest. The city adopted new rules on police use of facial recognition technology, but the ACLU, which represents Williams, argues the rules should become a new national standard.
“I'm hopeful this is a step in the right direction,” Williams said.
Williams was the first person known to be wrongfully arrested because of a facial recognition glitch, but he wouldn't be the last: Detroit police arrested at least two people because of failed facial recognition searches, including a woman who was charged with carjacking while eight months pregnant.
Law enforcement agencies across the country are using facial recognition technology to identify criminals caught on camera. In Michigan, the software compares unknown faces to those in a database of mugshots and driver's license photos. In other jurisdictions, police are using tools like Clearview AI, which searches photos harvested from social media sites and the public internet.
One of the most significant new rules introduced in Detroit is that images of people identified through facial recognition technology cannot be shown to witnesses in photo collages unless there is other evidence linking them to a crime.
“It ends the 'take a photo and put it on your mug shot,'” said Phil Meyer, an attorney with the ACLU of Michigan. “This settlement transforms the Detroit Police Department from one of the most documented police departments for misuse of facial recognition technology to a national leader in putting guardrails around its use.”
Police say facial recognition is a powerful tool for solving crimes, but some cities and states, including San Francisco, Austin, Texas, and Portland, Oregon, have temporarily banned its use over privacy and racial bias concerns. Steven Lamoreaux, director of intelligence for Detroit's Criminal Intelligence Division, said the police department is “very keen to use technology in a meaningful way for public safety,” and asserted that Detroit “has one of the strongest policies in the country right now.”
How the problem occurs
Williams was arrested for a crime that occurred in 2018, when a man was captured on surveillance camera stealing five watches from a boutique in downtown Detroit. An anti-theft company provided the footage to Detroit police.
According to documents unsealed in Williams' lawsuit, the system matched the man's face to driver's license photos and mugshots, finding 243 photos ranked by how confident it was that they matched the person in the security footage. Williams' old driver's license photo was ninth on the list. The person running the search determined that Williams was the best match and sent a report to a Detroit police detective.
Detectives put Williams' photo in a “six-person photo lineup” — a grid of six people — and showed it to the security contractor who provided the store's surveillance video. The contractor agreed that Williams was the closest match to the man at the boutique, which led to a warrant for his arrest. Williams, who was at his desk at the auto parts company when the watch was stolen, spent a night in jail and was fingerprinted and DNA-taken. He was charged with retail fraud and had to hire an attorney to defend himself. Prosecutors eventually dropped the case.
He sued Detroit in 2021, trying to force it to ban the technology to prevent others from experiencing what happened to him. He said he was outraged last year when he learned that a facial recognition mismatch led Detroit police to charge Porcha Woodruff with carjacking and robbery. Police arrested Woodruff as she was getting her children ready to go to school. She is also suing the city, and the case is ongoing.
Williams called facial recognition technology “very dangerous” and said he “doesn't see any positive benefit.”
New Rules
Detroit police are responsible for three of the seven known cases where facial recognition technology has led to wrongful arrests (the other two were in Louisiana, New Jersey, Maryland, and Texas), but Detroit officials say the new restrictions will prevent further misuse and are optimistic about the technology's potential to help solve crimes, and currently only use it for serious crimes like assault, murder, and burglary.
Detroit Police Chief James White blamed the wrongful arrests on “human error,” saying officers relied too heavily on clues generated by technology and that the errors were in the officers' judgment, not the machines.
The new policy, which goes into effect this month, should help: Under the new rules, police will no longer be able to show witnesses a person's face based solely on a facial recognition match.
“There has to be unrelated secondary corroborating evidence before there is sufficient justification for going into a lineup,” he said. Lamoreaux, of Detroit's Criminal Intelligence Unit. Police will likely need more than just physical similarities — location data from a suspect's cellphone, for example, or DNA testing.
Police are also changing the way they sort mugshots, using what's called a double-blind sequential method, which is considered a fairer method of identification. Instead of presenting witnesses with a “six-pack,” officers who don't know who the prime suspect is present each photo. The sorting also includes photos that differ from those of people flagged by the facial recognition system.
Police should also disclose that a facial search was conducted and the quality of the facial images searched (What was the resolution of the security cameras? How visible was the suspect's face?) since poor quality images make it difficult to get reliable results, the age of the photos displayed by the automated system, and whether there were other photos of the person in their database that did not match.
Detroit Deputy Police Chief Franklin Hayes said he believes the new measures will prevent future misidentifications.
“There are still some issues that could arise, like identical twins,” Hayes said. “You can't say for sure that there won't be any issues, but we think this is the best we've got so far.”
Alan Ross, a computer science professor at Michigan State University and an expert on facial recognition technology, said Detroit's policy is a great start and that other agencies should adopt it.
“We don't want to trample on individual rights or privacy, but we also don't want crime to become rampant,” Ross said.
How helpful is it?
Eyewitness identification is a difficult task, and police are turning to cameras and facial recognition technology as a more reliable tool than imperfect human memory.
White told local lawmakers last year that facial recognition “helped us take 16 murderers off our streets.” Asked for more details, police officials declined to provide details about the cases.
Instead, to demonstrate their technology, police officials played surveillance footage of a man pouring fuel into a gas station and setting it on fire. Police say the man was identified using facial recognition technology and arrested that night. He later pleaded guilty.
The Detroit Police Department is one of the few police departments that monitors facial recognition searches, submitting weekly reports to an oversight board about its use. Over the past few years, the department has averaged more than 100 searches a year, about half of which result in a potential match.
Police departments track only how often they find leads, not whether they turn up any fruit. But as part of the settlement, for which Williams also received $300,000, a police spokesman said, the department must conduct an audit of its facial-recognition searches going back to when it first began using the technology in 2017. If the department uncovers other cases of arrests with little or no supporting evidence beyond a facial match, it is expected to notify prosecutors.
Molly Kleinman, director of the University of Michigan's Technology Studies Center, said the new protections seem promising but she remains skeptical.
“Detroit is an incredibly surveilled city. There are cameras everywhere,” she says. “If this surveillance technology really worked as advertised, Detroit would be one of the safest cities in the country.”
Willie Barton, a member of the Police Commission, the oversight group that approved the new policy, said he remains opposed to police use of facial recognition technology but called it a “step in the right direction.”
“The technology is not ready yet,” Burton said. “One false arrest is one too many. Three in Detroit should be a wake-up call that we need to stop false arrests.”