Good morning. It’s Wednesday. Today we’ll look at concerns about the Police Department’s use of facial recognition technology.
The Police Department uses facial recognition technology thousands of times a year, but the wrongful arrest of a Brooklyn man in connection with a flashing incident has raised concerns about how effective the tool really is.
The man, Trevis Williams, was arrested in April after he was identified using facial recognition, even though he was about eight inches taller and 70 pounds heavier than the flasher described by the victim, my colleagues Maria Cramer and Kashmir Hill reported. The victim, a woman, picked Williams out of a photo lineup.
Williams had been driving from Connecticut to Brooklyn on the day of the crime, and location data from his phone put him 12 miles from the scene at the time when the woman, a cleaner on East 17th Street in Manhattan, said she had been flashed by an Amazon delivery man she recognized.
Williams spent two days in jail in April, and his case was ultimately dismissed.
“In the blink of an eye, your whole life could change,” Williams said.
The search for the flasher
After the flashing, the victim described the flasher to the police. (The New York Times is withholding her name because she was the victim of a sex crime.)
Police officers also canvassed the area and ran a facial recognition search using video from a surveillance camera, according to police documents. That search produced a match with Williams because his mug shot was in the system as a result of an assault charge that was ultimately dismissed.
When a facial recognition search is conducted, a person’s image is uploaded to a system that uses algorithms to render a face’s contours into data points and then looks for the faces that are the most statistically similar. A human then reviews the candidates and decides which ones appear to be the best matches.
The examiner chose Williams and generated a report that warned that he was only a “possible match,” which by itself was “not probable cause to arrest.”
His photo was then shown to the victim alongside pictures of six other people, all Black men with dreadlocks and facial hair. The woman picked Williams’s photo and said she was confident he was the flasher.
The investigation
Brad Weekes, a spokesman for the Police Department, said it was “factually inaccurate” to say that the police had made false arrests based on facial recognition technology.
But the case against Williams began to fall apart not long after he was arrested.
Investigators were confronted by the vast physical differences between him and the flasher, the cellphone data that placed him far from the crime and the fact that he had worked for Amazon but did not start the job until two months after the crime.
A spokeswoman for Amazon also said that investigators had never asked them for the name of the unidentified delivery man who had been accused of the crime in the first place. She said the company would have cooperated if it had been told of the investigation.
‘Only the tip of the iceberg’
Part of the problem may have also been the use of surveillance footage in making the match.
In 2023, the National Institute of Standards and Technology found that the leading face recognition algorithms were able to produce a correct match for 99.9 percent of searches they performed on a mug shot database of 12 million people.
But “accuracy drops sharply when low-quality or uncontrolled images are used,” researchers from the National Academies of Sciences, Engineering and Medicine said about that study. That includes images like those derived from surveillance video.
Legal Aid, the public defenders group that represented Williams, said it knew of other cases where the wrong person had been arrested after being falsely identified with facial recognition technology.
“We are gravely concerned that the cases we have identified are only the tip of the iceberg,” Legal Aid said in a letter it sent Monday to the city’s Department of Investigation, asking the agency to look into police use of the technology.
The American Civil Liberties Union has gone one step further and called for a ban on the use of facial recognition because of the risk of misidentification.
Weather
Expect a sunny day with temperatures in the high 70s. For tonight, expect a clear sky with temperatures in the low 60s.
ALTERNATE-SIDE PARKING
In effect until Monday (Labor Day).
The latest New York news
-
Keeping a watch on flooding in real time: The New York City Department of Environmental Protection, prompted by flood damage from Hurricane Ida, has invested in the tool FloodNet. City officials have installed more than 250 sensors around the city to track flooding.
-
Tisch says National Guard isn’t needed here: Police Commissioner Jessica Tisch told the U.S. attorney general, Pam Bondi, on Monday that crime is down in New York City and the National Guard is not needed, a person familiar with their conversation said. President Trump signed an executive order to create a specialized National Guard unit under his control.
-
Prosecutors recommend a seven-year sentence for Nadine Menendez: Menendez will face sentencing in September for her role in a bribery scheme involving her husband, former Senator Robert Menendez of New Jersey. Mr. Menendez is serving an 11-year term.
-
Former music director of Trinity church charged: Julian Wachner was charged in Indianapolis with possessing child sexual abuse material. He was fired three years ago from Trinity Church in Lower Manhattan amid an investigation into an allegation of sexual misconduct.
-
Mayoral candidates’ songs of the summer: Jon Caramanica, a Times critic, and Joe Coscarelli, a reporter, asked 10 cultural figures, including Zohran Mamdani and Curtis Sliwa, what they had been listening to. See their picks here.
-
New York’s Vietnamese food boom: For many people, cooking is a way to preserve culture and strengthen communities. New restaurants led by second-generation chefs are now offering innovative and authentic Vietnamese dishes in Manhattan.
METROPOLITAN diary
Salsa Savior
Dear Diary:
I was walking past the 34th Street ferry terminal when I noticed a young woman holding a jar of salsa in front of her and looking expectantly at the people streaming off an arriving boat.
I assumed she was looking for a particular person and, sure enough, I saw her run toward a young man in a white shirt who was walking in her direction.
Watching her run toward him, I expected to see a hug, a kiss or some other kind of physical greeting.
Instead, she held out the jar of salsa.
Moving a bit closer, I observed the man opening the jar and the woman jumping gleefully up and down.
“Thank you!” she exclaimed. “Thank you so much.”
Her mission apparently accomplished, she walked away from the man and carried the jar to a nearby bench, where a companion was waiting with a bag of chips.
— Eric Mathern
Illustrated by Agnes Lee. Tell us your New York story here and read more Metropolitan Diary here.
Glad we could get together here. — L.S.
P.S. Here’s today’s Mini Crossword and Spelling Bee. You can find all our puzzles here.
Francis Mateo, Luke Caramanico and Ed Shanahan contributed to New York Today. You can reach the team at [email protected].
Sign up here to get this newsletter in your inbox.
Liam Stack is a Times reporter who covers the culture and politics of the New York City region.
The post A Wrongful Arrest and Worry About the Accuracy of a Police Tool appeared first on New York Times.
