The New York Police Department (NYPD), known for its significant resources, with over 48,000 full-time staff and a budget approaching $6 billion, resembles a military force comparable to entire countries like the Philippines and Iraq. Unsurprisingly, the NYPD invests heavily in technology. Between 2007 and 2020, $2.8 billion was allocated to a range of surveillance technologies, including stingray phone trackers, crime prediction software, and X-ray vans.
One of the department’s most controversial investments is a facial recognition system developed since 2011, which recently resulted in the arrest of a man substantially different in appearance from the suspect. According to the New York Times, Trevis Williams was mistakenly arrested on April 21 after the facial recognition software flagged him as a probable match from grainy CCTV footage.
Police aimed to identify a man who had committed public lewdness in February. The algorithm generated six potential matches, all African American men with facial hair and dreadlocks. Despite an investigator warning that these matches were not probable cause, Williams’ photo was included in a lineup. The victim’s confident identification of him as the perpetrator gave police the grounds to mark him as a suspect.
While Williams was later apprehended in Brooklyn, the discrepancies in the suspect’s description versus Williams’ actual characteristics—being 12 miles away from the crime scene, eight inches taller, and 70 pounds heavier—did not prevent his arrest and two-day detention. Williams firmly denied the allegations, and although the charge was dismissed in July and the original investigation closed, the incident underscored the risks of unregulated use of technology in policing.
This case is not isolated. In Detroit, at least three other Black individuals have faced similar wrongful arrests due to facial recognition, leading to demands for strict policies on the use of such technology. The NYPD currently lacks these safeguards, and it is uncertain if and when they will be implemented.
