In January 2020, Robert Williams spent 30 hours in a Detroit jail after facial recognition technology suggested he was a criminal. The game went wrong, and Mr. Williams sued.
On Friday, as part of a legal settlement over wrongful arrest, Mr. Williams received a promise from the Detroit Police Department that it would do better. The city has adopted new rules for police use of facial recognition technology that the American Civil Liberties Union, which represents Mr. Williams, says should become a new national standard.
“We hope this moves the needle in the right direction,” Williams said.
Mr. Williams was the first person known to have been wrongfully arrested due to a false facial recognition search. But he was not the last. Detroit police have arrested at least two people as a result of a failed facial recognition search, including a woman who was accused of carjacking when she was eight months pregnant.
Law enforcement agencies across the country are using facial recognition technology to try to identify crimes caught on camera. In Michigan, the software compares unknown faces to faces in a database of mugshots or driver's license photos. In other jurisdictions, police use tools like Clearview AI to search for photos scraped from social media sites and the public internet.
One of the most significant new rules adopted in Detroit is that images of people identified through facial recognition technology can no longer be shown to witnesses in photo lineups unless there is other evidence linking them to a crime.
“This will end the ‘photo-and-lineup’ pipeline,” said Phil Mayor, staff attorney for the ACLU of Michigan. “This settlement transforms the Detroit Police Department from one of the most well-documented misusers of facial recognition technology to a national leader in using guardrails.”
Police say facial recognition technology is a powerful tool to help solve crimes. However, in some cities and states, including San Francisco; Austin, Texas; Portland, Oregon, temporarily banned its use due to concerns about privacy and racial bias. Stephen Lamoreaux, director of informatics for the Detroit Crime Intelligence Unit, said the police department is “very committed to using technology in meaningful ways for public safety.” He claimed Detroit has “the strongest policies in the country right now.”
How It Goes Wrong
Mr. Williams was arrested for a crime that occurred in 2018. A man stole five watches from a downtown Detroit boutique, and the incident was captured on surveillance camera. A loss prevention company provided the footage to the Detroit Police Department.
According to documents released as part of Mr. Williams’ lawsuit, a search of the man’s face against driver’s license photos and mugshots produced 243 photos, ranked by the system’s confidence that the person in the surveillance footage was the same person. Mr. Williams’ old driver’s license photo was ranked ninth on the list. The person conducting the search deemed him the best match and sent a report to Detroit police detectives.
Detectives included Mr. Williams' photo in a “six-pack photo lineup” – photos of six people arranged violently – that they showed to a security contractor who provided surveillance video of the store. She agreed that Mr. Williams was the closest person to the man in the boutique, which led to a warrant being issued for his arrest. Mr Williams, who was sitting at a desk at an auto supply company when his watch was stolen, spent a 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 dismissed the case.
He sued Detroit in 2021, hoping to force a ban on the technology to prevent others from suffering his fate. He said he was upset when he learned last year that Detroit police had charged Pocha Woodruff with carjacking and robbery because facial recognition matching wasn't working properly. Police arrested Woodruff as he was dropping his children off at school. She also sued the city, and the lawsuit is ongoing.
“It’s too risky,” Williams said, referring to facial recognition technology. “I don’t see any positive benefit in it.”
new rules
Detroit police are responsible for three out of seven cases of wrongful arrests due to facial recognition. (The others were in Louisiana, New Jersey, Maryland and Texas.) But Detroit officials said the new regulations will prevent more abuses. And they remain optimistic about the technology's crime-solving potential, currently only using it in cases of serious crimes, including assaults, murders and home invasions.
Detroit Police Chief James White blamed “human error” for the wrongful arrests. His officers, he said, rely too much on leads generated by technology. It was their judgment that was flawed, not the machine's.
A new policy that comes into effect this month is expected to help with this. Under the new rules, police can no longer show a person's face to witnesses based solely on a facial recognition match.
“There has to be some kind of secondary supporting evidence, even if it’s not relevant, before there’s sufficient justification to put it in a lineup,” he said. Mr. LaMoreaux of the Detroit Crime Intelligence Unit. Police will need more than location information or DNA evidence from a person's cell phone – a physical resemblance.
The department is also changing how it conducts photo lineups. We are adopting a double-blind sequential approach, which is considered a fairer way to identify someone. Instead of presenting a 'six pack' to a witness, a police officer who doesn't know who the main suspect is is presented with one photo at a time. And the lineup included a photo of a different person than what was shown in the facial recognition system.
Police must also disclose that face searches have taken place and the quality of the facial images being searched. How grainy were the surveillance cameras? How prominent is the suspect's face? — Because low-quality images are unlikely to produce reliable results. It should also be determined whether there are other photos in the database that do not match the age of the photo displayed in the automated system.
Detroit Police Department Deputy Chief Franklin Hayes said he is confident the new practice will prevent future misidentifications.
“There are still some mistakes you can make, for example with identical twins.” Mr. Hayes said. “I can’t say never, but I think this is the best policy so far.”
“Detroit’s policy is a great starting point,” said Arun Ross, a computer science professor at Michigan State University and an expert in facial recognition technology. “Other institutions should also adopt it.”
“We don’t want to trample on people’s rights and privacy, but we don’t want crime to run rampant,” Ross said.
How helpful is it?
Identifying witnesses is difficult, and police have embraced cameras and facial recognition as more reliable tools than the imperfect human memory.
Chief White told local lawmakers last year that facial recognition technology helped “take 16 murderers off the streets.” When asked for more information, police officials declined to provide details about the case.
Instead, police officials played surveillance footage of a man dousing a gas station with fuel and setting it on fire to show the department’s success with the technology. They said he was identified by 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 and submits weekly reports to the oversight committee on their use. Over the past few years, it has averaged more than 100 searches per year, with about half of those resulting in a potential match.
The department only tracks how often leads are turned up, not whether they are actually turned up. But as part of the settlement with Williams (who also received $300,000), the department must conduct an audit of facial recognition searches since it first began using the technology in 2017, according to a police spokeswoman. If it finds other instances where people have been arrested with little or no evidence other than a facial match, it must notify the relevant prosecutor.
Molly Kleinman, director of the University of Michigan's Center for Technology Research, said the new protections look promising but remain skeptical.
“Detroit is a very surveilled city. There are cameras everywhere,” she said. “If all this surveillance technology actually works as they claim, Detroit would be one of the safest cities in America.”
Willie Burton, a member of the police commission, the group of overseers who approved the new policy, described it as “a step in the right direction,” but he still opposed allowing police to use facial recognition technology.
“The technology is not ready yet,” Mr. Burton said. “One false arrest is one too many, and three in Detroit should be a wake-up call to stop it.”