Experts Talk Best Practices For Facial Recognition Technology San Francisco's move to ban facial recognition software has worried the industry, which would prefer regulation. But what should regulation look like? We talk to the experts on both sides.
NPR logo

Experts Talk Best Practices For Facial Recognition Technology

  • Download
  • <iframe src="https://www.npr.org/player/embed/724433238/724433264" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
Experts Talk Best Practices For Facial Recognition Technology

Experts Talk Best Practices For Facial Recognition Technology

Experts Talk Best Practices For Facial Recognition Technology

  • Download
  • <iframe src="https://www.npr.org/player/embed/724433238/724433264" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript

San Francisco's move to ban facial recognition software has worried the industry, which would prefer regulation. But what should regulation look like? We talk to the experts on both sides.

ARI SHAPIRO, HOST:

San Francisco's Board of Supervisors voted earlier this week to ban the city's use of facial recognition software. That's jump-started a debate about what rules might make the technology more acceptable to people. NPR's Martin Kaste has more.

MARTIN KASTE, BYLINE: Go somewhere that's already bristling with cameras - say, a neighborhood around a big marijuana store here in central Seattle - and ask people what they'd think if these cameras were hooked up to facial recognition software. You're mostly going to get reactions like this.

CARA BRAUN: I don't want my face to be recognized on the streets. I feel like it's my right to privacy. And if I want someone to know where I'm at and when I'm there, then I'll tell them.

KASTE: That's Cara Braun, definitely someone who likes the idea of a complete ban. Does that kind of attitude worry the industry?

BENJI HUTCHINSON: It does. It does. It worries us because, from the all-out simplistic ban perspective of the technology, it doesn't really seem like there's sufficient justification.

KASTE: Benji Hutchinson is VP of federal operations for NCE Corporation of America. It's one of the biggest sellers of facial recognition to law enforcement. And after this vote in San Francisco, he wants people to take a moment.

HUTCHINSON: There needs to be a discussion around the use cases of the technology and what specific instances are appropriate for using the technology. It needs to start there.

KASTE: He's right that facial recognition can be used in very different ways, and people tend to see some uses as worse than others. On one extreme is the blanket surveillance that China is trying to build - cameras everywhere capable of live tracking crowds of people. That use has few defenders here in America, at least in public. On the other end of the spectrum, though, is crime investigations, after the fact.

JAMES HART: Burglaries where a homeowner's surveillance camera captured a suspect coming up and breaking into their house.

KASTE: That's Detective James Hart with the Sacramento Sheriff's Department. He uses facial recognition in the way that's most typical in America right now - taking an image from a crime scene and trying to find a match. But right there is a big question for the reformers - whom should police be matching these images to? Just the people in their mug shot files, or everybody who's ever had a driver's license photo? Also, the reformers say cops should get a warrant first. Hart says if that became the rule, it would definitely change things for him.

HART: If it came down to the point where we were having to write search warrants on each one of these cases in order to use facial recognition, I mean, it would severely delay us in investigating additional crimes.

KASTE: Claire Garvie tracks law enforcement use of facial recognition for Georgetown Law's Center on Privacy and Technology. She'd like to see a complete moratorium. But if police are going to use this, she says it should be reserved for serious crimes.

CLAIRE GARVIE: And here's why - if the threshold is set too low, we could very quickly imagine a world where face recognition, like what's been trialled in China, is used to identify jaywalkers and send you a jaywalking traffic ticket, if you will.

KASTE: As to the other end of that spectrum of uses, the China style real-time street surveillance of crowds, she says there definitely have to be legal limits on that.

GARVIE: A limited location, limited time, a limited number of people that the system can identify, so that this is not a system that, just because there's somebody on FBI's most wanted list, the FBI gets to run face surveillance on every single camera across the country.

KASTE: But so far, limiting law enforcement's use of facial recognition is not something that Congress seems very interested in. There is a bill to regulate how companies use it. Brian Schatz, the senator from Hawaii, is one of the authors.

BRIAN SCHATZ: I think this issue is all brand new to most members of Congress. And so we're starting with the commercial sector because we think there's a better chance to get consensus.

KASTE: At the state level, too, lawmakers have been paying more attention to commercial users than law enforcement. And this suspicion about the private sector is echoed by Detective Hart in Sacramento. Even though he uses facial recognition in his work, he doesn't like the idea of stores using it on him.

HART: I don't want to be tracked myself, specifically also to give some sort of edge to a business who's trying to make more money.

KASTE: As to expanding law enforcement's uses, Hart says he can see using live facial recognition at, say, an airport. But he thinks the average person should still be able to, in his words, walk around and have his own personal life be his own personal life. Martin Kaste, NPR News.

(SOUNDBITE OF ANGEL OLSEN'S "THOSE WERE THE DAYS")

Copyright © 2019 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.