SALT LAKE CITY — San Francisco recently became the first city in the United States to ban the use of facial recognition software by police and other city departments.
Though generally considered the nation’s most tech-centric area, the city said this ban on technology is part of broader legislation that requires departments to establish use policies and obtain approval before deploying surveillance technology.
The move by San Francisco is also part of a larger conversation and controversy happening around privacy. Last year, civil rights groups called on Amazon CEO Jeff Bezos to stop selling the company’s facial recognition technology to the government, and others claim this type of surveillance feels a little too close to Big Brother.
Facial recognition tech is a fairly new field, and its implications are just starting to be felt and understood. So what’s happening in Utah, another rapidly-expanding, tech-centric area?
It turns out there aren’t many city or state agencies using the tech just yet. One of the few exceptions is Utah’s Driver License Division in the Department of Public Safety. The agency uses the tech mainly for fraud prevention, according to division director Chris Caras.
The software can identify characteristics of a person’s face — things like facial structure or hair color — then match that person’s face to similar or identical faces in the system. Every night, the division runs all the pictures taken during the day to see “if somebody has applied for identification under your name or vice versa,” Caras said.
“We reconcile whether they are the same record or the same person on record, or whether it indicates whether there are multiple records with the same individual involved,” he added. “We catch a lot more up front that way.”
When the department first initiated this procedure in January 2010, they went back through old files and caught a lot of fraudulent records as they “cleaned house,” Caras said. Most of those records were things like fake IDs for underage drinkers, but fraud has since become a little more sophisticated, he explained.
The false records the division catches now are often created for financial scams or other more nefarious purposes. There’s definitely less of it, but it’s more refined, he said.
Luckily for the victims of these crimes, the division has a wait period before an applicant’s information is sent to the driver license manufacturer. If the system catches a fraudulent record, the division can rectify the issue before any false record is physically created.
Caras explained, however, that the division does not use its facial recognition technology in conjunction with law enforcement to, say, identify a possible suspect caught on Walmart surveillance footage.
While there are software systems that are able to inject, scan and match photos from “outside” feeds (like a surveillance tape), the division chose to use a system that can only match the pictures in its system with other pictures in its system.
According to Caras, the division uses this latter system for a couple of reasons.
“We take privacy very seriously and we don’t want to violate the public trust. We know that people come to us seeking a service and that we provide a credential that doesn’t just say they have the privilege to operate their motor vehicle but is also the most widely-used identification documentation. So we take that very seriously, but we also know they came to us with that in mind. They weren’t meaning to facilitate a lot of other functionality, and so we try to be respectful of that.”
It’s “kind of nice” when other agencies occasionally ask the division for help identifying someone to simply say the system can’t handle it, Caras said. If the division ever does decide to upgrade their system to something that can inject outside feed, a discussion about privacy is definitely something that will need to happen, he added.
Both the Salt Lake City Police Department and the Unified Police Department say they do not employ any facial recognition technology, though UPD spokeswoman Sgt. Melody Gray said her detectives wish they had access to it.
Salt Lake detective Greg Wilking said their department has had “conversations about having conversations” regarding the tech, but the real question is whether facial recognition would be worth the cost. The department would use it mainly to scan faces and identify potential known threats during large events, but it would be difficult to match surveillance footage to a face unless it was exceptionally good surveillance footage, he said.
Even then, an individual’s right to privacy may present a roadblock. Caras is still unsure how a state should walk the line between public safety and public privacy but believes that's an issue that may quickly become relevant.
“I would hate for us as a licensing, credentialing, identification kind of entity to lose (facial recognition) as a tool because we have seen a real benefit in being able to combat people who are making fraudulent applications for nefarious reasons, and I believe that we’ve been able to protect people’s identity at a much higher level having that functionality,” he said. “(But), to respect the public privacy, there has to be a line.”