From the U.S. Government Accountability Office, www.gao.gov Transcript for: Expanded Use of Facial Recognition Technology has Increased Privacy and Accuracy Concerns Description: The use of facial recognition technology to identify an individual has increased in recent years--leading to growing concerns about how it impacts privacy, and how accurate it is. We discuss with GAO's Alicia Puente Cackley, a director in our Financial Markets and Community Investment team, who has a new report on facial recognition technology. Related GAO Work: GAO-20-522, Facial Recognition Technology: Privacy and Accuracy Issues Related to Commercial Uses Released: August 2020 [Intro Music] [Alicia Puente Cackley:] Some of the privacy concerns are things like loss of anonymity in public spaces and also a lack of knowledge about how the technology is being used. [Holly Hobbs:] Hi, and welcome to GAO's Watchdog Report, your source for news and information from the U.S. Government Accountability Office. I'm Holly Hobbs. The use of facial recognition technology to verify or identify an individual has increased in recent years. Most recently, some companies are using this kind of technology to monitor the spread of the coronavirus. While facial recognition technology has its benefits, there are growing concerns about how it impacts privacy and how accurate it is. Today, we talked to Alicia Puente Cackley, a director in our Financial Markets and Community Investment team, who has a new report on facial recognition technology. Thank you for joining us, Alicia. [Alicia Puente Cackley:] Thank you for having me. [Holly Hobbs:] So, Alicia, how are companies currently using facial recognition technology? [Alicia Puente Cackley:] So, right now a key commercial use of facial recognition technology is to confirm a person's identity, so providing access to a building or an event or even a car. It can be used to control access to a smartphone or a bank account instead of using a password, which is good for preventing fraud. And then, some retailers and event venues are using facial recognition technology for safety and security purposes, to deter theft or to identify banned individuals. [Holly Hobbs:] And how has this use evolved over time? [Alicia Puente Cackley:] Well, in the beginning, the use of the technology was for marketing and customer service, but that's really expanded even more in recent years. So, for example, now it's used by retailers to do very, targeted VIP marketing to VIP customers. Hotels can use it to facilitate check-in processes, sort of a no-touch process. And then, newer uses are things like authorizing payments and also tracking attendance. So, schools or businesses could use it to track attendance of students and employees. [Holly Hobbs:] And how are companies using this technology to monitor the spread of the coronavirus? [Alicia Puente Cackley:] Some companies are using the technology to identify individuals who may have been exposed to the coronavirus. So, for example, one company's website describes using first thermal-imaging cameras to measure a building's occupant's body temperature, and then they use facial recognition technology to identify who may have come into contact with any of those individuals who displayed a fever. [Music] [Holly Hobbs:] So, it sounds like facial recognition technology is not only being used more often, but in more ways, which seems like it could impact people's privacy. Alicia, what are the privacy concerns about facial-recognition technology and what's the federal role in protecting privacy? [Alicia Puente Cackley:] Well, some of the privacy concerns are things like loss of anonymity in public spaces and also a lack of knowledge about how the technology is being used. So, groups we spoke with highlighted concerns about whether companies were adequately providing consumers with clear notice about how their images were being used; and also around whether those companies were receiving consent from those consumers for the images to be used. And then, there's development of facial-recognition data sets, so large groups of photos. We heard concerns about companies that use web-scraping techniques to create their data sets. So, that's automatic software that extracts data from different websites. And if that's done without consent of the individuals in the images, then that's a concern. And unfortunately, the federal role in protecting privacy is pretty limited around this technology. In most contexts, federal law doesn't address how personal data derived from facial recognition technology can be used or shared, and while there are some existing federal laws that cover specific types of data, like financial data or health data, and there is laws that protect personal information for children under the age of 13, there is no comprehensive federal privacy law. [Holly Hobbs:] And your report also discusses accuracy concerns. Can you explain what you found? [Alicia Puente Cackley:] Overall, we found that facial recognition accuracy has increased dramatically in recent years, mostly through the use of artificial intelligence, but despite that increase in accuracy, we found that there are performance issues that continue to exist between different demographic groups. So, in general, the technology performs better on white men than on black women. For example, in many cases, the technology was between 10 and 100 times more accurate on white men than on black women, when it came to incorrectly matching two faces that are actually from two different people. The technology also generally had worse performance on other minorities, as well as on children and the elderly. And at the same time, a small number of developers have produced facial-recognition technology with no detectable performance differences between demographic groups. So, there's quite a variation. And the differences in accuracy matter because they could result in more frequent misidentification for certain demographics, such as misidentifying a shoplifter. It could result in somebody being blocked from access to something that they legitimately should have access to, because they've been misidentified. [Holly Hobbs:] Alicia, last question. What is the bottom line of this report? [Alicia Puente Cackley:] Our bottom line is that the market for facial recognition technology has increased and it's likely to continue to expand as a touchless form of verifying and identifying individuals. But, without federal legislation in place to address privacy, accuracy, and data-security issues, concerns related to the technology will likely persist. [Holly Hobbs:] That was Alicia Puente Cackley discussing GAO's recent review of facial-recognition technology. Thank you for your time, Alicia. [Alicia Puente Cackley:] You're very welcome. [Holly Hobbs:] And thank you for listening to the Watchdog Report. To hear more podcasts, subscribe to us on Apple Podcasts and make sure you leave a rating and review to let others know about the work we're doing. For more from the congressional watchdog, the U.S. Government Accountability Office, visit us at gao.gov.