PITTSBURGH — The potential use of face-recognition technology in a massive surveillance network set up by the Allegheny County district attorney has raised alarms among community groups and the state chapter of the American Civil Liberties Union.
In a letter to District Attorney Stephen A. Zappala Jr. late last week, 21 signatories asked the county’s top prosecutor to “commit to not using facial recognition technology and to having greater transparency with respect to use of surveillance technology.”
“The use of facial recognition as a law enforcement tactic sends a message that the only way to keep us safe is by treating us as threats to be monitored, tracked, and incarcerated, using ever-more-sophisticated technology,” the letter, signed by Pennsylvania ACLU executive director Reginald T. Shuford.
Zappala’s spokesman, Mike Manko, declined to comment until Shuford has a chance to see Zappala’s response, which should be coming “within the next couple of days.”
Zappala has spent about $1.5 million assembling a 1,000-camera surveillance network that spans five counties in southwestern Pennsylvania.
An investigation by The Caucus, a watchdog publication of LNP Media Group, uncovered details about the network that shocked privacy and national security experts. Those include the use of a secret private contractor to oversee the network, lack of written agreements governing access to video feeds, and the presence of Chinese-made components so vulnerable to foreign and domestic hacking that the U.S. Congress designated them national security threats last year.
Facial recognition capabilities are not currently active in the network, Manko told The Caucus. Many of the cameras use automated license-plate-recognition technology that can track vehicles in real time or reconstruct its movements over several weeks. Manko and Zappala have credited the cameras with helping solve bank robberies, thefts and kidnappings, and finding two murder suspects.
One of Zappala’s deputies told a community forum in Pittsburgh earlier this year that the use of facial recognition technology in the network is coming “down the road.” The aid, Dick Skrinjar, told the group that law enforcement agents could upload photos of public school students who are on probation so the system could help them “restrict where these people go, and keep them out of areas they’re not supposed to be in.”
Studies by M.I.T. and the U.S. Department of Commerce, among others, have shown facial recognition technology more accurately identifies white faces than those with darker skin. The ACLU ran the faces of members of Congress through Amazon’s facial recognition system, called Rekognition, and it falsely matched 28 of them with mugshots, according to the letter.
“Use of facial recognition technology would disparately impact people of color and women, and fray relationships and trust between law enforcement and police communities,” according to Shuford’s letter.
He called the idea of using facial recognition to monitor school children “particularly troubling.”
“Facial recognition technology is less accurate with children’s faces, and children should not be subjected to surveillance,” Shuford wrote. “Children are still developing and should be given opportunities for treatment, rehabilitation, and positive reinforcement.”