United States Government Accountability Office Report to Congressional Requesters August 2021 FACIAL RECOGNITION TECHNOLOGY Current and Planned Uses by Federal Agencies GAO-21-526 August 2021 FACIAL RECOGNITION TECHNOLOGY Current and Planned Uses by Federal Agencies Highlights of GAO-21-526 a report to congressional requesters Why GAO Did This Study What GAO Found Facial recognition—a type of biometric technology—mimics how people identify or verify others by examining their faces Recent advancements have increased the accuracy of automated FRT resulting in increased use across a range of applications As the use of FRT continues to expand it has become increasingly important to understand its use across the federal government in a comprehensive way In response to GAO’s survey about facial recognition technology FRT activities in fiscal year 2020 18 of the 24 surveyed agencies reported using an FRT system for one or more purposes including GAO was asked to review the extent of FRT use across the federal government This report identifies and describes 1 how agencies used FRT in fiscal year 2020 including any related research and development and interactions with non-federal entities and 2 how agencies plan to expand their use of FRT through fiscal year 2023 GAO surveyed the 24 agencies of the Chief Financial Officers Act of 1990 as amended regarding their use of facial recognition technology GAO also interviewed agency officials and reviewed documents such as system descriptions and information provided by agencies that reported using the technology • Digital access or cybersecurity Sixteen agencies reported using FRT for digital access or cybersecurity purposes Of these 14 agencies authorized personnel to use FRT to unlock their agency-issued smartphones—the most common purpose of FRT reported Two agencies also reported testing FRT to verify identities of persons accessing government websites • Domestic law enforcement Six agencies reported using FRT to generate leads in criminal investigations such as identifying a person of interest by comparing their image against mugshots In some cases agencies identify crime victims such as exploited children by using commercial systems that compare against publicly available images such as from social media • Physical security Five agencies reported using FRT to monitor or surveil locations to determine if an individual is present such as someone on a watchlist or to control access to a building or facility For example an agency used it to monitor live video for persons on watchlists and to alert security personnel to these persons without needing to memorize them Ten agencies reported FRT-related research and development For example agencies reported researching FRT’s ability to identify individuals wearing masks during the COVID-19 pandemic and to detect image manipulation Furthermore ten agencies reported plans to expand their use of FRT through fiscal year 2023 For example an agency plans to pilot the use of FRT to automate the identity verification process at airports for travelers Examples of Facial Recognition Technology Uses by Federal Agencies View GAO-21-526 For more information contact Candice N Wright at 202 512-6888 or wrightc@gao gov or Gretta L Goodwin at 202 512-8777 or goodwing@gao gov United States Government Accountability Office Contents Letter 1 Background Agencies Most Often Reported Using FRT for Digital Access and Domestic Law Enforcement Ten Agencies Plan to Expand Use of FRT Mostly through Use of New FRT Systems Agency Comments 3 25 29 Appendix I Objectives Scope and Methodology 31 Appendix II Summaries of Selected Federal Agencies’ Facial Recognition Technology Activities 38 Appendix III Comments from the U S Agency for International Development 81 Appendix IV Comments from the Social Security Administration 83 Appendix V GAO Contacts and Staff Acknowledgments 84 9 Tables Table 1 Facial Recognition Technology FRT Activities Reported by Federal Agencies for Fiscal Year 2020 Table 2 Reported Purposes of Facial Recognition Technology Systems Used by Federal Agencies in Fiscal Year 2020 Table 3 Federally Owned Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 Table 4 Commercially Owned Facial Recognition Technology FRT Systems Accessed by Federal Agencies in Fiscal Year 2020 Table 5 Federal Agencies That Reported Plans to Expand Their Use of Facial Recognition Technology FRT Systems through Fiscal Year 2023 Page i 9 11 16 20 25 GAO-21-526 Facial Recognition Technology Table 6 New Facial Recognition Technology FRT Systems Federal Agencies Reported they Plan to Use through Fiscal Year 2023 26 Figures Figure 1 Process used in Facial Recognition Technology Figure 2 Examples of Federal Agencies’ Use of Facial Recognition Technology FRT Figure 3 States and Localities that Own Facial Recognition Technology FRT Systems Accessed by Federal Agencies in Fiscal Year 2020 Page ii 5 8 18 GAO-21-526 Facial Recognition Technology Abbreviations CBP CFO Act COVID-19 DHS DOD DOE DOJ DOT EPA FACE FBI FRT GSA HHS HSIN IDENT NASA NCRFRILS NSF OPM PIN RAPIDS SSA TSA USAID USDA VA U S Customs and Border Protection Chief Financial Officers Act of 1990 Coronavirus Disease 2019 Department of Homeland Security Department of Defense Department of Energy Department of Justice Department of Transportation Environmental Protection Agency Facial Analysis Comparison and Evaluation Federal Bureau of Investigation facial recognition technology General Services Administration Department of Health and Human Services Homeland Security Information Network Automated Biometric Identification System National Aeronautics and Space Administration National Capital Region Facial Recognition Investigative Leads System National Science Foundation Office of Personnel Management personal identification number Real-time Automated Personnel Identification System Social Security Administration Transportation Security Administration U S Agency for International Development U S Department of Agriculture Department of Veterans Affairs This is a work of the U S government and is not subject to copyright protection in the United States The published product may be reproduced and distributed in its entirety without further permission from GAO However because this work may contain copyrighted images or other material permission from the copyright holder may be necessary if you wish to reproduce this material separately Page iii GAO-21-526 Facial Recognition Technology Letter 441 G St N W Washington DC 20548 August 24 2021 The Honorable Jim Jordan Ranking Member Committee on the Judiciary House of Representatives The Honorable Carolyn B Maloney Chairwoman The Honorable James Comer Ranking Member Committee on Oversight and Reform House of Representatives Of all the biometric technologies—those used to identify people based on their biological and behavioral characteristics—facial recognition most closely mimics how people identify others by examining their faces What is an effortless skill in humans has proven difficult to replicate in machines but computer and technology advancements over the past few decades have increased the overall accuracy of automated facial recognition As a result the use of facial recognition technology FRT has become increasingly common across business and government sectors For example it is used as a tool for identifying or verifying customers and to verify an employee’s identity when logging into a computer Law enforcement can also use it to search databases such as driver’s license photos and mugshots for possible leads about an unknown individual’s identity as part of a criminal investigation As the use of FRT continues to expand Members of Congress academics and advocacy organizations have highlighted the importance of developing a comprehensive understanding of how it is used by federal agencies This report is the latest in a series of recent reports we have issued on FRT In June 2021 we reported on federal law enforcement’s use of FRT including the extent of its use and how the agencies monitor such use 1 In September 2020 we reported on the U S Customs and 1GAO Facial Recognition Technology Federal Law Enforcement Agencies Should Better Assess Privacy and Other Risks GAO-21-518 Washington D C June 3 2021 In this report we made 26 recommendations related to using nonfederal systems with facial recognition technology to eight agencies including the Departments of Health and Human Services Homeland Security the Interior Justice State and the Treasury Page 1 GAO-21-526 Facial Recognition Technology Border Protection’s CBP and Transportation Security Administration’s TSA use of FRT at U S ports of entry and made recommendations to CBP to improve its privacy practices and system performance 2 In 2016 we reported on the Federal Bureau of Investigation’s FBI use of FRT and made recommendations to improve the Bureau’s understanding of the accuracy of and privacy protection processes for its FRT capabilities 3 We have also recently reported on the ways FRT can be used in commercial settings including to provide secure access to online customer accounts and information on customer flows during peak times among others 4 You asked us to review the extent of FRT use across the federal government This report identifies and describes 1 how agencies used FRT in fiscal year 2020 including any FRT-related research and development activities and interactions with nonfederal entities and 2 how agencies plan to expand their use of FRT through fiscal year 2023 To address these objectives we administered a survey to the 24 Chief Financial Officers CFO Act agencies 5 to collect information related to 1 FRT systems used owned or accessed including those tested by 2GAO Facial Recognition CBP and TSA are Taking Steps to Implement Programs but CBP Should Address Privacy and System Performance Issues GAO-20-568 Washington D C Sept 2 2020 In this report we made five recommendations to CBP related to its use of facial recognition technology and DHS concurred with the recommendations In March and April 2021 CBP provided a status update on progress towards each of these recommendations Based on the documentation provided by CBP GAO closed two recommendations as implemented 3GAO Face Recognition Technology FBI Should Better Ensure Privacy and Accuracy GAO-16-267 Washington D C May 16 2016 In this report we made six recommendations related to accuracy and privacy The FBI has addressed all six recommendations 4GAO Facial Recognition Technology Privacy and Accuracy Issues Related to Commercial Uses GAO-20-522 Washington D C July 13 2020 and GAO Facial Recognition Technology Commercial Uses Privacy Issues and Applicable Federal Law GAO-15-621 Washington D C July 30 2015 5The 24 agencies are those identified in the Chief Financial Officers Act of 1990 as amended 31 U S C § 901 b They are the U S Departments of Agriculture Commerce Defense Education Energy Health and Human Services Homeland Security Housing and Urban Development the Interior Justice Labor State Transportation the Treasury Veterans Affairs Environmental Protection Agency National Aeronautics and Space Administration U S Agency for International Development General Services Administration National Science Foundation Nuclear Regulatory Commission Office of Personnel Management Small Business Administration and Social Security Administration Page 2 GAO-21-526 Facial Recognition Technology agencies during fiscal year 2020 2 FRT systems agencies planned to use through fiscal year 2023 3 agencies’ FRT-related research and development activities 4 transactions financial or otherwise that agencies entered into for nonfederal entities’ use of FRT and 5 the extent to which agencies regulated nonfederal entities’ use of FRT 6 The questionnaire also asked detailed questions about the individual FRT systems that agencies reported which included the purpose s a brief description of its use and obligations related to its use We asked agencies to include the activities of all their components bureaus and offices in their responses We emailed questionnaires to the agencies in October 2020 and closed the survey in January 2021 after receiving responses from all 24 agencies We reviewed the responses we collected and took quality control steps by performing checks for completeness logical errors and inconsistencies We followed up with agencies in writing or through interviews as appropriate See appendix I for additional information on our scope and methodology We conducted this performance audit from April 2020 through August 2021 in accordance with generally accepted government auditing standards Those standards require that we plan and perform the audit to obtain sufficient appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives We believe that the evidence obtained provides a reasonable basis for our findings and conclusions on our audit objectives Background How Facial Recognition Technology Works Facial recognition can verify or identify individuals by their faces It is one of several biometric technologies that identify individuals by measuring and analyzing physical and behavioral characteristics 7 As seen in figure 1 facial recognition technology uses a photo or still from a video feed of a person—often called a probe or live photo—and converts it into a template or a mathematical representation of the photo A matching 6Regulated refers to using regulatory authority over a nonfederal entity to regulate that entity’s use of its own FRT For the purposes of our questionnaire we defined “regulated” as regulatory functions in which the agency engaged including but not limited to investigatory and inspections activities taking enforcement actions prescribing requirements or guidance conducting oversight and maintaining performance standards 7Other biometric technologies can identify individuals by measuring and analyzing physical and behavioral characteristics which include fingerprints eye irises voice and gait Page 3 GAO-21-526 Facial Recognition Technology algorithm can then compare the template to one from another photo and calculate their similarity Facial recognition searches or comparisons generally fall into two categories verification and identification Verification or one-to-one searches compares a stored photo of an individual to another photo purportedly of the same individual to determine whether they are the same person For example this type of comparison can help verify the identity of an individual attempting to unlock a smartphone Identification or one-to-many searches compares a photo from a single individual against a gallery of stored photos from a number of individuals to determine if there is a potential match For example this type of comparison can be used to identify investigative leads for an unknown individual in a crime scene photo Page 4 GAO-21-526 Facial Recognition Technology Figure 1 Process used in Facial Recognition Technology Page 5 GAO-21-526 Facial Recognition Technology Two technologies facial detection and facial analysis are related to but distinct from facial recognition Whereas facial recognition matches a face to a specific identity • Facial detection determines if a photo or video contains a face in the image It is commonly used to count the number of people that move through a particular area without determining their identities such as counting people in stores or amusement park lines • Facial analysis sometimes referred to as facial classification or characterization uses a facial image to estimate or classify personal characteristics such as age race or sex or tracks facial features or movement to recognize expressions or gaze among other analyses For example facial analysis can be part of an eye tracking system which can allow researchers to analyze how well pilots use their eyes or gaze to scan their cockpit instruments For the purposes of this report we use the term “facial recognition technology” to include facial recognition facial detection or facial analysis technologies Federal Use of Facial Recognition Technology Systems A facial recognition technology system may include components or modules of systems software applications or devices with automated facial recognition capabilities such as a face recognition algorithm hardware or software Federal agencies can own their FRT systems or access the FRT systems of other government entities including federal state local tribal and territorial governments and commercial facial recognition service providers 8 Agencies can have direct access to an FRT system such as by logging into the system or indirect access such as by requesting a state government i e a third party run a facial recognition search on behalf of the federal agency FRT systems are used for a variety of purposes across the federal government These purposes and examples of how FRT can be used are grouped into seven different categories as follows 9 8We use the term “federally owned” when an agency developed or acquired the FRT and performs its own searches We use the term “commercially owned” when an agency contracts for an FRT service such as a search performed against the commercial entity’s database of images or performed by the commercial entity itself and the results are provided back to the agency 9For the purposes of this report we determined these categories to describe the different ways agencies could use FRT Page 6 GAO-21-526 Facial Recognition Technology • Digital access or cybersecurity This purpose includes FRT that can be used to control access to a personal computer smartphone or mobile application • Domestic law enforcement This purpose includes FRT that can be used to identify a lead or person of interest in an investigation or to locate or identify a missing person or crime victim • Physical security This purpose includes FRT that can be used to control physical access such as to facilities or buildings or to surveil or monitor a location or facility including notification that an individual is present in real-time • Border and transportation security This purpose includes FRT that can be used to confirm the identities of domestic travelers at airports travelers applying to enter the United States or crossing U S borders or non-citizens in immigration proceedings • National security and defense This purpose includes FRT that can be used to research derogatory information on a known or suspected terrorist or confirm the identity of a foreign national for national security reasons 10 • Medical assessment This purpose includes FRT that can be used to confirm a patient’s identity in a medical setting such as when dispensing controlled substance prescriptions or to assist with contact tracing e g related to Coronavirus Disease 2019 or COVID19 11 • Other These FRT purposes include analysis of the face itself such as analysis of attention or alertness based on eye tracking or inferring characteristics of a person including age or sex It also includes any other agency use that does not fit into the categories above Illustrative examples of these purposes are shown in figure 2 10For example to confirm the identity of a foreign national seeking access to a military installation 11For information on how contact tracing can occur see GAO Science Tech Spotlight Contact Tracing Apps GAO-20-666SP Washington D C Jul 28 2020 Page 7 GAO-21-526 Facial Recognition Technology Figure 2 Examples of Federal Agencies’ Use of Facial Recognition Technology FRT Page 8 GAO-21-526 Facial Recognition Technology Agencies Most Often Reported Using FRT for Digital Access and Domestic Law Enforcement Most of the federal agencies we surveyed—19 of 24—reported one or more FRT-related activities in fiscal year 2020 with digital access and domestic law enforcement as the most common For the purposes of our survey we identified four types of FRT activities an agency can engage in 1 using an FRT system which includes owning accessing or testing the system 2 conducting or supporting FRT-related research and development 3 entering into transactions with nonfederal entities such as awarding grants to enable entities to obtain FRT systems for their own uses and 4 regulating the use of FRT by nonfederal entities Table 1 shows the four types of FRT activities reported in fiscal year 2020 by the 24 federal agencies we surveyed Five agencies reported they did not conduct any of these FRT activities in fiscal year 2020 the Department of Education Department of Housing and Urban Development Department of Labor the Nuclear Regulatory Commission and the Small Business Administration Table 1 Facial Recognition Technology FRT Activities Reported by Federal Agencies for Fiscal Year 2020 Federal Agency Used FRT systemsa ● ● ● Department of Agriculture Department of Commerce Department of Defense Department of Education ⊗ ● ● ● Department of Energy Department of Health and Human Services Department of Homeland Security Department of Housing and Urban Development Department of the Interior Department of Justice Department of Labor Department of Transportation Department of Veterans Affairs U S Agency for International Development Environmental Protection Agency General Services Administration National Aeronautics and Space Administration Page 9 ⊗ ⊗ ⊗ ⊗ ● ● ⊗ ● ● ⊗ ⊗ ⊗ Department of the Treasury ⊗ ⊗ ● ● ● ● ● ● ● ● Regulated nonfederal entities’ use of FRTd ⊗ ⊗ ● Department of State Entered into Conducted FRTtransactions related research with nonfederal and developmentb entities for FRTc ⊗ ● ● ● ⊗ ● ⊗ ⊗ ⊗ ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ● ●e ⊗ ⊗ ⊗ ⊗ ● ⊗ ● ⊗ ⊗ ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ GAO-21-526 Facial Recognition Technology Federal Agency Used FRT systemsa ● National Science Foundation Nuclear Regulatory Commission ⊗ ● Office of Personnel Management Small Business Administration ⊗ ● Social Security Administration ● Yes ⊗ No Entered into Conducted FRTtransactions related research with nonfederal and developmentb entities for FRTc ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ Regulated nonfederal entities’ use of FRTd ⊗ ⊗ ⊗ ⊗ Source GAO analysis of survey results GAO-21-526 The agency owned or accessed FRT in fiscal year 2020 a Research and development also includes agencies that obligated funds for FRT-related research conducted by other entities b c Refers to agencies that entered into transactions such as awarding grants to nonfederal entities to purchase FRT equipment Refers to using regulatory authority over a nonfederal entity to regulate that entity’s use of its own FRT For the purposes of our questionnaire we defined “regulated” as regulatory functions in which the agency engaged including but not limited to investigatory and inspections activities taking enforcement actions prescribing requirements or guidance conducting oversight and maintaining performance standards d e The Transportation Security Administration within the Department of Homeland Security used its authority to regulate entities under its jurisdiction Eighteen Agencies Reported Using FRT for a Variety of Purposes Eighteen of the 24 agencies we surveyed responded that they used facial recognition technology in fiscal year 2020 as shown in the first column of table 1 above 12 These agencies reported using FRT for one or more purposes with digital access and domestic law enforcement as the most common see table 2 13 Agencies did not report using FRT for medical assessment purposes in fiscal year 2020 12In our questionnaire we asked “At any point in fiscal year 2020 did you agency use facial recognition technology for any of the following purposes ” These purposes are 1 digital access or cybersecurity 2 domestic law enforcement 3 physical security 4 border and transportation security 5 national security and defense 6 medical assessment and 7 other purposes For the purposes of this questionnaire and report “use” refers to whether an agency 1 owned and or operated a FRT system 2 accessed directly or through a third party an FRT system as part of a program or activity within their agency but that was owned by another federal or nonfederal entity or 3 tested a FRT system as part of a pilot proof of concept trial or evaluation for potential agency use 13Agencies Page 10 reported that some FRT systems were used for multiple purpose categories GAO-21-526 Facial Recognition Technology Table 2 Reported Purposes of Facial Recognition Technology Systems Used by Federal Agencies in Fiscal Year 2020 Purpose Federal Agency Digital access ● ● Department of Agriculture Department of Commerce Department of Defense ⊗ ● ● ● ● ● Department of Energy Department of Health and Human Services Department of Homeland Security Department of the Interior Department of Justice Department of State ⊗ Department of the Treasury Department of Veterans Affairs Agency for International Development Environmental Protection Agency General Services Administration National Aeronautics and Space Administration National Science Foundation Office of Personnel Management Social Security Administration ● Yes ⊗ No ● ● ● ● ● ● ● ● ● National security and defense Other Domestic law enforcement Physical security Border and transportation security ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ● ⊗ ● ● ● ● ⊗ ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ● ● ● ● ⊗ ⊗ ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ● ⊗ ⊗ ● ⊗ ● ⊗ ● ⊗ ● ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ● ⊗ ⊗ ⊗ ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ● ⊗ ⊗ ⊗ Source GAO analysis of survey results GAO-21-526 Note Agencies did not report using FRT for medical purposes in fiscal year 2020 Page 11 GAO-21-526 Facial Recognition Technology Examples of how agencies used FRT in each purpose category are described below • Digital access or cybersecurity Sixteen agencies reported using FRT for digital access or cybersecurity purposes 14 Of these 14 agencies authorized personnel to use FRT to unlock their agencyissued smartphones—the most common purpose of FRT reported by the agencies in our survey 15 Two agencies—General Services Administration GSA and Social Security Administration SSA — reported conducting pilots that used agency employees to test FRT systems as a means to control access to certain government websites such as GSA’s login gov 16 Specifically GSA and SSA used FRT to compare two images—a government photo identification and a live image of the individual—to verify the identity of an individual attempting to apply for an account This FRT system may also conduct a check to detect if there is an attempt to subvert the FRT using a printed image or other non-live object 17 However agency officials said that this FRT would not be deployed until additional testing under a range of conditions is completed 14GAO’s cybersecurity work encompasses a broad range of issues assessing information security as a government-wide high-risk area including protecting cyber critical infrastructure and protecting the privacy of personally identifiable information This body of work also addresses agency compliance with federal cybersecurity requirements and includes assessments of security controls As used in this report cybersecurity refers to the general framework that includes digital access and other controls 15The 14 agencies that reported using facial recognition to unlock smartphones or tablets are the Departments of Agriculture USDA Commerce Homeland Security DHS Energy DOE Justice DOJ Health and Human Services HHS the Interior the Treasury Veterans Affairs VA the Environmental Protection Agency EPA the National Aeronautics and Space Administration NASA the National Science Foundation NSF the Office of Personnel Management OPM and the U S Agency for International Development USAID Many of these agencies reported that they did not require FRT when procuring their smartphones rather it was a feature that was built into the smartphone We did not inquire whether an agency monitored an employee’s use of facial recognition to unlock their agency-issued smartphones or had policies related to its use However OPM reported that they no longer authorized personnel to use the feature 16Login gov is a publicly accessible website that verifies the identities of individuals seeking to access participating agencies’ websites 17This check detects whether a facial recognition system sensor is viewing data from a live subject as opposed to recorded data of a non-living object For example a picture or a 3D mask or a printed image that may be presented to try to fool the FRT system into false authentication Page 12 GAO-21-526 Facial Recognition Technology • Domestic law enforcement Six agencies—Departments of Homeland Security DHS Justice DOJ Defense DOD Health and Human Services HHS the Interior and the Treasury—reported using FRT to generate leads in criminal investigations such as identifying a person of interest by comparing images of the person against databases of mugshots or from other law enforcement encounters 18 For example DOJ’s Federal Bureau of Investigation used the Next Generation Identification Interstate Photo System to generate leads during investigations by comparing photos of unknown individuals suspected of criminal activity against a repository of photos of known individuals including mugshots and other records In addition agencies may access commercially owned FRT systems as part of criminal investigations or to assist in identifying a missing person or victims of crimes such as exploited children For example DHS DOJ HHS and the Interior reported using Clearview AI a commercially owned facial recognition system that compares a submitted photo against a database of publicly available images from open sources such as social media and returns matching images for review • Physical security Five agencies—Department of Commerce DOD Department of Energy DOE DOJ and HHS—reported using FRT to monitor or surveil locations to determine if an individual is present such as someone from a watchlist or to control access to a building or facility For example HHS reported that it used an FRT system AnyVision to monitor its facilities by searching live camera feeds in real-time for individuals on watchlists or suspected of criminal activity which reduces the need for security guards to memorize these individuals’ faces This system automatically alerts personnel when an individual on a watchlist is present In addition DOJ reported using an FRT system to verify that personnel attempting entry into their on-site secure network operations centers at federal prisons were authorized for entry • Border and transportation security Two agencies—DHS and Department of State—reported using FRT systems to assist with 18DOD’s criminal investigative organizations e g Naval Criminal Investigative Service and Air Force Office of Special Investigations and other DOD law enforcement agencies or organizations e g military police departments can use FRT for domestic law enforcement in addition to other purposes when “that information logically relates to the detection neutralization or deterrence of criminal activity that affects DOD personnel property or mission ” See Department of Defense DOD Instruction No 5505 17 Collection Maintenance Use and Dissemination of Personally Identifiable Information and Law Enforcement Information by DOD Law Enforcement Activities Washington D C Dec 19 2012 rev Nov 29 2016 Page 13 GAO-21-526 Facial Recognition Technology identifying or verifying travelers within or seeking admission to the United States identifying or verifying the identity of non-U S citizens already in the United States and to research agency information about non-U S citizens seeking admission to the United States For example DHS’s U S Customs and Border Protection used its Traveler Verification Service at ports of entry to assist with verifying travelers’ identities The Traveler Verification Service uses FRT to compare a photo taken of the traveler at a port of entry with existing photos in DHS holdings which include photographs from U S passports U S visas and other travel documents as well as photographs from previous DHS encounters 19 • National security and defense Four agencies—DHS DOD DOJ and State—reported using FRT for national security and defense purposes including to identify individuals known or suspected to be terrorists research derogatory information about a suspected threat actor and monitor or surveil locations to search for a person of interest such as a suspected terrorist For example the State Department reported using the Integrated Biometric System which uses FRT to perform searches of visa and passport applicants’ photos against terrorist watchlist photos The Integrated Biometric System provides consular posts and passport agencies around the world with additional information to evaluate visa and passport applications to decrease the possibility that a terrorist would be able to fraudulently receive a U S visa or passport • Other purposes Three agencies—DOD DOJ and the National Aeronautics and Space Administration NASA —reported using FRT for other purposes including to verify the identities of individuals receiving identification cards and temporary badges and demonstrate how FRT works in educational settings For example DOD tested a facial detection capability to support issuance of DOD identification cards within its Real-time Automated Personnel Identification System RAPIDS Specifically RAPIDS uses facial detection when capturing the cardholder’s picture during initial enrollment or identification card issuance or renewal to ensure that the size of the picture printed on the cards is consistent Similarly NASA’s Johnson Space Center reported testing a prototype FRT system to confirm an employee’s identity by comparing a current camera image of the employee with a photo on file 20 19See GAO-20-568 for more detail 20NASA reported that it conducted the experiment for a limited time only and did not continue working on or using the prototype after fiscal year 2020 Page 14 GAO-21-526 Facial Recognition Technology Eighteen Agencies Reported Owning and Accessing Facial Recognition Technology According to our analysis of survey responses 18 agencies reported they owned FRT systems or accessed other entities’ FRT systems in fiscal year 2020 Of these agencies 17 owned or accessed federal FRT systems three accessed FRT systems owned by state and local entities and six accessed FRT systems owned by commercial vendors Seventeen Agencies Reported Owning or Accessing Federal FRT Systems Seventeen agencies reported they owned or accessed 27 federal FRT systems 21 Fourteen of these agencies owned smartphones that can be unlocked with facial recognition and three of these agencies—U S Agency for International Development EPA and OPM—did not own or access any other FRT systems 22 Nine agencies—Commerce DOD DOE HHS DHS DOJ State GSA and NASA—owned FRT systems other than smartphones as shown in table 3 Three of these agencies— DHS DOD and DOJ—owned 18 of the 27 federal FRT systems in addition to owning smartphones Finally one agency—Treasury—did not own an FRT system but accessed federal and commercially owned systems 23 21In some cases agencies reported their systems as owned at the component-level or below and that access may be given to other components within the same agency We report such cases as a single instance of ownership by the agency Appendix II provides more detailed descriptions of the federal FRT systems including how agencies use them whether they were accessed by other agencies and more information on the systems 22We only counted smartphones that can be unlocked using facial recognition as a single FRT system even though there are multiple vendors offering similar technologies on their smartphones Fourteen agencies reported they owned smartphones that can be unlocked with facial recognition These agencies include USDA Commerce DOE HHS DOJ DHS Interior Treasury VA USAID EPA NASA NSF and OPM 23Treasury reported it accessed the General Services Administration’s login gov during testing of the FRT capability Login gov is a single-sign on mechanism that uses FRT to match applicants to their identification documents to access accounts on agency websites as mentioned earlier Treasury also reported a third-party vendor performed facial recognition searches on its behalf in April 2020 Page 15 GAO-21-526 Facial Recognition Technology Table 3 Federally Owned Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 Agency Number of FRT Systems Owned Department of Commerce 1 system Accessed FRT Systems Owned by Other Agencies • None Department of Defense 7 systems • • Department of Justice Department of Homeland Security Purposes • Physical security • Physical security Domestic law enforcement National security and defense Other Physical security • • • Department of Energy 1 system • None • Department of Health and Human Services 3 systems • None • • • Department of Homeland Security 4 systems • • • Department of Defense Department of Justice Department of State • • • Department of Justice 7 systems • • Department of Defense Department of State • • • • Department of State 1 system • Department of Defense • • General Services Administration 1 system • None • National Aeronautics and Space Administration 1 system • None • Physical security Domestic law enforcement Digital access or cyber security Domestic law enforcement Border and transportation security National security and defense Domestic law enforcement Physical security National security and defense Other Border and transportation security National security and defense Digital access or cyber security Other Source GAO analysis of survey results GAO-21-526 Note The table excludes agencies that only reported owning smartphones that can be unlocked with facial recognition See appendix II for additional information about agency use of FRT systems Of the nine agencies that owned an FRT system other than smartphones five agencies—Commerce DOE GSA HHS and NASA—reported they owned federal FRT systems and did not access other federal FRT systems For example HHS tested an FRT system that agency personnel could use to unlock their laptops NASA tested another FRT system that verified an employee’s identity by comparing a camera image with a photo on file if the employee forgot their badge Commerce and DOE owned FRT systems that controlled personnel access to secure facilities Page 16 GAO-21-526 Facial Recognition Technology For example DOE’s TacID Guard Dog performs facial matching and facial verification from live video to monitor the entry and exit of agency personnel at access points The other four agencies—DOD DHS DOJ and State—reported that they owned an FRT system and also accessed FRT systems owned by other federal agencies For example the State Department’s Bureau of Consular Affairs reported owning the Integrated Biometric System which is also accessed by DHS and DOJ State Department officials used this FRT system to verify a visa applicant’s identity or determine whether an individual has previously applied for a visa under an alias Similarly DHS accessed the Integrated Biometric System to verify the identity of individuals applying for visas and immigration benefits and to identify individuals being investigated for identity theft and benefits fraud In addition the State Department’s Bureau of Diplomatic Security reported accessing DOD’s Automated Biometric Identification System to verify the authenticity of an individual’s travel documents among other things Three Agencies Reported Accessing FRT Systems Owned by State and Local Entities Three agencies—DOJ DHS and Interior—reported accessing one or more FRT systems owned by 29 states and seven localities for law enforcement purposes Figure 3 shows the states and localities that own FRT systems accessed by these federal agencies Page 17 GAO-21-526 Facial Recognition Technology Figure 3 States and Localities that Own Facial Recognition Technology FRT Systems Accessed by Federal Agencies in Fiscal Year 2020 DOJ and DHS reported direct and indirect access to FRT systems across states and localities 24 For example DOJ reported access to FRT systems through personnel in the FBI’s Facial Analysis Comparison and Evaluation FACE Services which has memoranda of understanding with 21 states and two federal entities to access their FRT systems 24The term ‘direct access’ refers to cases where federal agency personnel can log into an FRT system and perform a facial recognition search The term ‘indirect access’ refers to cases where federal agency personnel request that the owner of an FRT system conduct a facial recognition search Page 18 GAO-21-526 Facial Recognition Technology Specifically FBI agents can submit a probe photo and request FACE Services examiners perform facial recognition searches for their investigations FACE Services examiners have direct access to one state FRT system and indirect access to FRT systems owned by 20 other state entities 25 In addition DHS reported it owns the Homeland Security Information Network HSIN which enables access to the Multi-State Facial Recognition Community of Interest While HSIN is not an FRT system it has a form for authorized users to request indirect facial recognition searches through state and local entities such as fusion centers 26 Through various memoranda of understanding DHS personnel in U S Customs and Border Protection and U S Immigration and Customs Enforcement may submit photos for facial recognition searches to those entities These searches assist DHS personnel with identifying individuals involved in identity theft and benefit fraud investigations and targets connected to criminal investigations or to a known terrorist organization In fiscal year 2020 DHS officials reported the agency made facial recognition search requests to 15 state and four local partner agencies including state and local law enforcement fusion centers Finally Interior reported it accessed the National Capital Region Facial Recognition Investigative Leads System NCRFRILS NCRFRILS is an FRT system that contains copies of information including photos from participating law enforcement agencies in the Washington D C metro area Interior’s U S Park Police reported that in July 2020 it requested the Maryland National Capital Park Police perform a facial recognition search of NCRFRILS on its behalf to generate investigative leads 25According to FBI officials for direct access to the Maryland Department of Public Safety and Correctional Services’ FRT system examiners have completed related training requirements so a memorandum of understanding is unnecessary FACE Services has direct access to two federal FRT systems—the FBI’s Next Generation Identification Interstate Photo System and the Department of State’s visa holdings The FBI can request the Department of State perform facial recognition searches on passport holdings on the FBI’s behalf and return only a limited number of photos that are likely matches 26HSIN is for trusted sharing of Sensitive but Unclassified information—such as that related to law enforcement or homeland security—between federal state local territorial tribal international and private sector partners Generally fusion centers are collaborative information-sharing efforts to detect prevent investigate and respond to potential criminal activity including terrorism Page 19 GAO-21-526 Facial Recognition Technology Six Agencies Accessed Commercial FRT Systems Six agencies—HHS DHS Interior DOJ Treasury and SSA—reported accessing eight FRT systems owned by commercial vendors Table 4 Commercially Owned Facial Recognition Technology FRT Systems Accessed by Federal Agencies in Fiscal Year 2020 Number of FRT Systems Accessed Commercial Vendor Department of Health and Human Services 1 system • Clearview AI • Domestic law enforcement Department of Homeland Security 2 systems • Clearview AI Vigilant Solutions • Agency Purposes Department of the Interior 1 system • Clearview AI • Domestic law enforcement Border and transportation security National security and defense Domestic law enforcement Department of Justice 2 systems • • Domestic law enforcement • Domestic law enforcement • Digital access or cybersecurity • • • Department of the Treasury 1 system • Clearview AI Vigilant Solutions Other Social Security Administration 1 system • Acuant FaceID • Source GAO analysis of survey results GAO-21-526 Note See appendix II for additional information about agency use of FRT systems As shown in table 4 four agencies—HHS DHS Interior and DOJ— accessed Clearview AI which conducts facial recognition searches using publicly available images 27 For example HHS’s Office of the Inspector General reported it began a pilot of Clearview AI in September 2020 to assist with identifying subjects of a criminal investigation Similarly DHS reported the U S Secret Service and U S Immigration and Customs Enforcement piloted Clearview AI in fiscal year 2020 to identify individuals in federal criminal investigations and to identify perpetrators and victims in domestic and international child exploitation cases respectively DHS also reported that since June 2019 U S Customs and Border Protection agents had direct albeit temporary access to a Clearview AI through their participation with the New York State Intelligence Center These agents used Clearview AI to identify criminals as well as subjects who have been arrested previously were deported and attempted to re-enter the United States at the border 27Interior’s U S Park Police reported that it stopped using Clearview AI in June 2020 after conducting a pilot test in April 2020 Page 20 GAO-21-526 Facial Recognition Technology Ten Agencies Reported Conducting or Supporting FRT-related Research and Development Based on our analysis 10 of the 24 agencies surveyed—Commerce DHS DOD DOJ the Department of Transportation DOT HHS VA NASA the National Science Foundation NSF and State—conducted or supported research and development R D for FRT in fiscal year 2020 28 Of these agencies DOT did not report any other FRT-related activities Four agencies—DHS DOD DOJ and State—generally focused their R D on agency-specific needs such as to develop new applications or improve existing capabilities Examples from each of these agencies include • DHS reported sponsoring Biometric Technology Rallies which are ongoing industry challenges to develop innovative solutions for biometric collection and matching including facial recognition For example the 2020 Rally focused on the ability of FRT systems to reliably collect or match images of individuals wearing masks DHS intended this research to improve the technology’s ability to recognize individuals without requiring them to remove their protective equipment • DOD reported researching new capabilities for RAPIDS which would support identity verification during online identification card renewal and PIN reset requests • DOJ reported conducting applied research on the relationship between skin tone and false match rates in facial recognition algorithms the capabilities and limitations of current synthetic face detection such as deepfakes and the development of software to detect synthetic faces 29 DOJ also explored the potential benefits of combining FRT systems with trained forensic examiners to achieve better matching performance than by the technology or by humans alone • The State Department reported conducting research and development and contributing to international image standards for travel documents For example State conducted research on morphing detection and the impact of aging on the accuracy of facial recognition 28In our questionnaire we asked “In fiscal year 2020 did your agency conduct research and development R D involving facial recognition technologies ” 29For more information on deepfakes see GAO Science Tech Spotlight Deepfakes GAO-20-379SP Washington D C Feb 20 2020 Page 21 GAO-21-526 Facial Recognition Technology algorithms such as for children’s passport photos 30 State is also developing the ability to match images of individuals against passport images and a repository of known or suspected terrorists using the Personal Identification Secure Comparison and Evaluation System 31 Two agencies—Commerce and NSF—conducted or supported FRTrelated research more broadly including for commercial vendors and other agencies For example Commerce reported that its National Institute of Standards and Technology performed research to support the development of standards and methods in performance measurement image quality testing and evaluating technologies and interoperability for facial recognition technology The National Institute of Standards and Technology also conducted research through the Face Recognition Vendor Test program which most recently released reports quantifying facial recognition accuracy with facemasks using post-COVID-19 algorithms and across demographic effects 32 The NSF reported that it awards grants to universities and others to conduct research on facial recognition Specifically NSF’s Directorate for Computer and Information Science and Engineering supported FRTrelated research including a project assessing how to prevent identifying an individual from facial images used in research such as recordings of a driver’s face during driver behavior studies Furthermore NSF supported a program called the Center for Identification Technology Research in which university partners work with government and industry stakeholders on biometrics including research on FRT 33 Finally four agencies—DOT HHS the Department of Veterans Affairs VA and NASA—reported using FRT as a tool to conduct other research For example DOT reported that the Federal Railroad Administration used eye tracking to study alertness in train operators 30Face morphing is when for example two images of different individuals are combined so that the resulting image could be used as identification for both of them such as on a passport Morphing detection attempts to identify these images 31The State Department’s Bureau of Counterterrorism offers the Personal Identification Secure Comparison and Evaluation System or PISCES for border management under the Terrorist Interdiction Program to foreign partners and incorporates enhanced screening technologies to ensure those partners are able to protect themselves from attempts by terrorists to enter transit or depart their country 32DHS and DOJ have interagency agreements with the National Institute of Standards and Technology for related FRT research and evaluation 33DHS also sponsored the Center for Identification Technology Research Page 22 GAO-21-526 Facial Recognition Technology Like DOT NASA also reported that it used eye tracking to conduct human factors research In addition VA reported it used eye tracking as part of a clinical research program that treats post-traumatic stress disorder in veterans Specifically the eye tracking system evaluates pupil response to evaluate impairment Similarly HHS’s National Institutes of Health awarded grants for research that use eye tracking as a tool for clinical research For example characterizing how children with and without autism spectrum disorder visually followed conversations in videos among other areas of research Four Agencies Reported Other FRT-Related Activities with Nonfederal Entities According to our analysis of survey responses four agencies—DHS DOJ VA and State—reported FRT-related activities with nonfederal entities in fiscal year 2020 Specifically these agencies reported entering into transactions that enabled nonfederal entities to obtain their own FRT 34 These transactions included agreements with foreign governments and commercial entities grants and medical equipment purchases In addition DHS was the only agency that reported it regulated an airline’s use of FRT in fiscal year 2020 35 Transactions with Nonfederal Entities to Obtain FRT Agreements Two agencies—DHS and State—reported entering into FRT-related agreements in fiscal year 2020 with foreign governments and DHS reported having FRT-related agreements with commercial entities Specifically DHS entered into agreements called project arrangements with two foreign governments—Australia and the United Kingdom—related to the assessment of facial recognition software The State Department reported transactions with two foreign governments Specifically State contracted trainers to instruct Mexican government personnel on how to use previously donated FRT equipment and it purchased FRT equipment to donate to the Guatemalan government DHS also had cooperative research and development agreements with two commercial entities focused on making digital identity credentials 34In our questionnaire we asked “In fiscal year 2020 did your agency enter into transactions to enable a nonfederal entity to obtain facial recognition technology for their own uses In other words the support financial or in-kind would enable nonfederal entities to develop purchase or use facial recognition technology for their own uses—not for your agency’s use ” For the purposes of this questionnaire and report “transactions” refers to an agency that awarded grants entered into contracts leases or cooperative agreements provided direct loans or loan guarantees or entered into any other transactions with nonfederal entities using other transactional authority 35The Transportation Security Administration within the Department of Homeland Security regulated an airline’s use of FRT in fiscal year 2020 Page 23 GAO-21-526 Facial Recognition Technology e g digital driver’s licenses interoperable with airport checkpoint security systems Contract DHS’s U S Immigration and Customs Enforcement reported that it awarded a contract to the Lehigh County Pennsylvania District Attorney’s Office to enhance U S Immigration and Customs Enforcement’s future access to the Gang Intelligence Application which is a database of transnational gang members and associated information Grants Two agencies—DHS and DOJ—reported awarding FRT-related grants to nonfederal entities Specifically DHS’s Federal Emergency Management Agency awarded preparedness grants to sustain or build facial recognition and other capabilities such as intelligence sharing among state and local law enforcement emergency management and other local entities 36 DOJ reported it awarded a grant to the Police Foundation for the development of techniques to automate analysis of body-worn camera audio and video data of police and community interactions For example these techniques could allow for an evaluation of officers’ adherence to principles of procedural justice Medical equipment VA reported two FRT-related transactions for nonfederal entities Specifically VA purchased two types of eye tracking equipment for veterans According to the VA speech-impaired veterans were provided a prosthetic device that tracks eye movements to assist their use of a computer or tablet for communication Regulating Nonfederal Entities’ Use of FRT Of the 24 agencies in our survey one agency—DHS—reported regulating the use of FRT by other entities in fiscal year 2020 37 Specifically the TSA issued security program amendments to aircraft operators that 36DHS officials provided a list of the following preparedness grant programs that included “FRT” or “facial recognition” in the project description for awards in fiscal year 2020 Nonprofit Security Grant Program-Urban Area State Homeland Security Program and Urban Area Security Initiative 37In our questionnaire we asked “In fiscal year 2020 did your agency engage in any regulatory functions over nonfederal entities that use facial recognition technology ” Regulated refers to using regulatory authority over a nonfederal entity to regulate that entity’s use of its own FRT For the purposes of this questionnaire and report we defined “regulated” as regulatory functions in which the agency engaged including but not limited to investigatory and inspections activities taking enforcement actions prescribing requirements or guidance conducting oversight and maintaining performance standards Page 24 GAO-21-526 Facial Recognition Technology permit the use of FRT to identify passengers checking baggage for transportation on flights in lieu of the standard passenger identification measures in the applicable TSA-issued security program 38 Ten Agencies Plan to Expand Use of FRT Mostly through Use of New FRT Systems According to our analysis of survey responses 10 of the 24 agencies surveyed—USDA Commerce DOD HHS DHS Interior DOJ State Treasury and VA—plan to expand their use of FRT systems in one or more ways through fiscal year 2023 39 We categorized plans to expand FRT use in three ways 1 using new FRT systems 2 evaluating existing FRT systems e g pilot testing and 3 upgrading existing FRT systems 40 See table 5 for additional information Table 5 Federal Agencies That Reported Plans to Expand Their Use of Facial Recognition Technology FRT Systems through Fiscal Year 2023 Plan to use new FRT systems Federal Agency ● ● Department of Agriculture Department of Commerce Plan to evaluate FRT systems Plan to upgrade FRT systems or capabilities ⊗ ⊗ ⊗ ⊗ 38The TSA responded that it has broad authority to ensure transportation security See for example 49 U S C §§ 114 44901 44903 including research and development of new technologies 49 U S C § 44912 Specific to biometrics Congress authorized the TSA to use “voice stress analysis biometric or other technologies to prevent a person who might pose a danger to air safety or security from boarding the aircraft of an air carrier or foreign air carrier in air transportation or intrastate air transportation ” Aviation and Transportation Security Act of 2001 Pub L No 107-71 §109 a 7 115 Stat 597 2001 codified at 49 U S C § 114 note 39In our questionnaire we asked “Does your agency have plans to begin using facial recognition technology including upgrading a system to include facial recognition for internal agency use between fiscal year 2020 and fiscal year 2023 In other words your agency has not yet begun to use the facial recognition technology but it has taken steps to begin by fiscal year 2023 ” For the purposes of this questionnaire and report by “plans ” we meant that the agency has initiated a process to use facial recognition technology which could include an ongoing acquisition process a contract with a vendor or another agency a memorandum of understanding or a budget request It did not include hypothetical or exploratory conversations about potential uses of facial recognition technology within the agency Furthermore we did not ask agencies to confirm if they were planning to continue using their existing FRT systems beyond fiscal year 2020 However we included existing FRT systems in the planned use section when an agency reported plans to change the way it will use the FRT system from fiscal year 2020 through fiscal year 2023 40For the purposes of this report “upgrades” refers to cases where agencies reported adding FRT capabilities to existing systems or to enhance system processes and updating a previously deactivated FRT system It is not typical system maintenance Page 25 GAO-21-526 Facial Recognition Technology Plan to use new FRT systems Federal Agency ● ● ● ● ● ● ● ● Department of Defense Department of Health and Human Services Department of Homeland Security Department of the Interior Department of Justice Department of State Department of the Treasury Department of Veterans Affairs ● Yes Plan to evaluate FRT systems ● Plan to upgrade FRT systems or capabilities ⊗ ⊗ ⊗ ● ● ⊗ ⊗ ⊗ ⊗ ⊗ ⊗ ● ⊗ ⊗ ⊗ No ⊗ Source GAO analysis of survey results GAO-21-526 Using New FRT Systems Ten agencies—USDA Commerce DOD HHS DHS Interior DOJ State Treasury and VA—reported plans to use 17 new FRT systems through fiscal year 2023 as shown in table 6 41 New FRT systems refers to systems that are new to federal agencies including newly-developed FRT systems and commercial-off-the-shelf systems and new access to existing FRT systems that agencies did not report using in fiscal year 2020 According to our analysis of survey results agencies reported that 13 of the 17 new FRT systems will be owned by federal agencies and two by local governments Two agencies reported they plan to access Clearview AI a commercial system for the first time 42 Table 6 New Facial Recognition Technology FRT Systems Federal Agencies Reported they Plan to Use through Fiscal Year 2023 Agency Planned Number of Newly Owned and Accessed FRT Systems Planned Access to FRT Systems Owned by Other Purposes Agencies Department of Agriculture 2 systems • None Department of Commerce 1 system • None • • 41See • Physical security Domestic law enforcement Physical security appendix II for additional information about agency use of FRT systems 42Multiple Page 26 agencies already accessed Clearview AI in fiscal year 2020 noted above GAO-21-526 Facial Recognition Technology Agency Planned Number of Newly Owned and Accessed FRT Systems Planned Access to FRT Systems Owned by Other Purposes Agencies Department of Defense 2 systems • None • • • Physical security Domestic law enforcement National security and defense Department of Health and Human Services 1 system • None • Domestic law enforcement Department of Homeland Security 2 systems • None • • Domestic law enforcement National security and defense Department of the Interior 2 systems • None • Domestic law enforcement Department of Justice 2 systems • None • Physical security Border and transportation security Border and transportation security Domestic law enforcement • Department of State 1 system • None • Department of the Treasury 2 systems • Department of Defense Department of Homeland Security Department of Justice None • • • Department of Veterans Affairs 2 systems • • • Physical security Domestic law enforcement Source GAO analysis of survey results GAO-21-526 Note See appendix II for additional information about agency use of FRT systems Use of new federal FRT systems Nine agencies—USDA Commerce DOD HHS DHS DOJ State Treasury and VA—plan to use new federal FRT systems For example the U S Treasury Inspector General for Tax Administration reported that it purchased an FRT system that can identify facial images of persons of interest who may be involved in criminal activity across multiple investigations in December 2020 The FRT system searches images in an online storage locker which contains evidence such as photos from seized mobile devices and will notify investigators of potential matches of individuals linked to other investigations The State Department reported plans for a pilot in late 2021 using FRT developed for the Personal Identification Secure Comparison and Evaluation System PISCES border management system State plans to screen individuals against passport images and a repository of suspicious Page 27 GAO-21-526 Facial Recognition Technology individuals such as known and suspected terrorists attempting to travel through partner countries New access to existing FRT systems Three agencies—DOD Interior and Treasury—reported plans to access existing FRT systems 43 For example DOD’s U S Air Force Office of Special Investigations reported it began an operational pilot using Clearview AI in June 2020 which supports the agency’s counterterrorism counterintelligence and criminal investigations The agency reported it already collects facial images with mobile devices to search national databases and plans to enhance searches by accessing Clearview AI’s large repository of facial images from open sources to search for matches Evaluating Existing Systems Three agencies—DHS DOD and Treasury—reported plans to conduct new pilot tests or continue evaluating existing FRT systems Of the four FRT systems these agencies plan to evaluate federal agencies own three systems and a commercial vendor owns the other system For example DHS reported plans to initiate a new pilot and continue an ongoing pilot of an existing FRT system As of March 2021 the TSA is collaborating with U S Customs and Border Protection and a commercial airline at the Detroit Metropolitan Wayne County Airport to evaluate the use of biometric technology including facial recognition to automate the identity verification process at TSA checkpoints and streamline traveler experience In addition DOD and Treasury reported plans to conduct new pilots of existing FRT systems For example DOD plans to conduct a pilot in late fiscal year 2021 of an FRT enhancement to an electronic physical access control FRT system called Automated Installation Entry to improve processing and minimize security risks 44 DOD personnel that volunteer for the new pilot will proceed to the enhanced access control points which will match their faces against a database of DOD participants Upgrading Existing Systems DHS reported plans to upgrade an existing FRT system and capabilities through fiscal year 2023 In December 2021 DHS plans to replace IDENT which is its current system for processing and storing biometric data with the Homeland Advanced Recognition Technology system 43These systems are already used by other agencies 44The U S Army purchased the FRT system in fiscal year 2014 to verify individuals seeking access to military installations Page 28 GAO-21-526 Facial Recognition Technology Initially the Homeland Advanced Recognition Technology system will replace IDENT’s current capabilities to include FRT with DHS planning additional capabilities in subsequent years Agency Comments We provided a draft of this report to the 24 CFO Act agencies for their review and comment We received written comments from USAID and SSA that are reprinted in appendices III and IV respectively USAID in its written comments did not comment on the content of the report SSA in its written comments noted that the report was accurate with respect to their experience with facial recognition technologies We received technical comments from six agencies which we incorporated as appropriate We did not receive comments from the Department of the Treasury’s offices but received technical comments from some of its bureaus and components The remaining 15 agencies informed us that they had no comments We are sending copies of this report to the appropriate congressional committees the heads of the 24 CFO Act agencies and other interested parties In addition the report is available at no charge on the GAO website at https www gao gov Page 29 GAO-21-526 Facial Recognition Technology If you or your staff have any questions about this report please contact Candice N Wright at 202 512-6888 or wrightc@gao gov or Gretta L Goodwin at 202 512-8777 or goodwing@gao gov Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report GAO staff who made key contributions to this report are listed in appendix V Candice N Wright Director Science Technology Assessment and Analytics Gretta L Goodwin Director Homeland Security and Justice Page 30 GAO-21-526 Facial Recognition Technology Appendix I Objectives Scope and Methodology Appendix I Objectives Scope and Methodology This report identifies and describes 1 how agencies used facial recognition technologies FRT in fiscal year 2020 including any FRTrelated research and development activities and interactions with nonfederal entities and 2 how agencies plan to expand their use of FRT through fiscal year 2023 To obtain the information needed for both objectives we conducted a survey of the 24 agencies listed in the Chief Financial Officers CFO Act of 1990 as amended 1 These departments and independent agencies hereafter referred to as agencies are as follows • Department of Agriculture • Department of Commerce • Department of Defense • Department of Education • Department of Energy • Department of Health and Human Services • Department of Homeland Security • Department of Housing and Urban Development • Department of the Interior • Department of Justice • Department of Labor • Department of State • Department of Transportation • Department of the Treasury • Department of Veterans Affairs • Agency for International Development • Environmental Protection Agency • General Services Administration • National Aeronautics and Space Administration 1The 24 agencies are those identified in the Chief Financial Officers Act of 1990 as amended 31 U S C § 901 b This Act does not include many independent agencies and commissions such as the Office of the Director of National Intelligence or the Central Intelligence Agency so they were not included in our survey Page 31 GAO-21-526 Facial Recognition Technology Appendix I Objectives Scope and Methodology • National Science Foundation • Nuclear Regulatory Commission • Office of Personnel Management • Small Business Administration • Social Security Administration We administered a questionnaire by email to each of the 24 CFO Act agencies from October 2020 through January 2021 2 During survey administration we asked agencies to provide responses for all their components bureaus and offices in a consolidated response We received responses from all 24 agencies 3 To develop the questionnaire we used information from prior GAO reports and early interviews with the agencies to determine the areas of inquiry For the purposes of our survey and report we defined facial recognition technology as systems components or modules of systems software applications or devices with automated facial recognition capabilities such as face recognition algorithm hardware or software Facial recognition generally refers to facial matching which includes both verification one-to-one matching —to automatically confirm whether a facial image in one photo matches a facial image in a different photo— and identification one-to-many matching —to automatically determine whether a facial image has any match in a database or gallery of photos Though they are generally considered distinct technologies we also considered facial analysis—identifying attributes about a person based on their face such as sex age or emotion—and facial detection— determining if a photo or video contains a face— to be facial recognition technologies We also determined that facial recognition technology could be used for a variety of applications such as verifying the identity claimed by an individual or controlling access to buildings or computers We grouped these applications into seven purposes based on prior GAO reports such as those on law enforcement transportation security and commercial 2These dates cover the initial response for all 24 agencies It does not include updated or additional responses received because of follow-up activities 3We also asked agencies to include the 24 Offices of Inspectors General and the Treasury Inspector General for Tax Administration in their response Page 32 GAO-21-526 Facial Recognition Technology Appendix I Objectives Scope and Methodology uses of facial recognition and a review of relevant literature 4 We asked agencies to include all uses of facial recognition technology and identify which purpose s described their uses The purposes are 1 physical security 2 digital access or cybersecurity 3 domestic law enforcement 4 border and transportation security 5 national security and defense 6 medical assessment and 7 other purposes not already listed Finally we met with both agency liaisons who would be administering the survey to their respective agencies and subject matter experts who helped ensure respondents clearly understood specific questions We used a questionnaire with two sections 1 a Main Questionnaire and 2 five Attachments that followed up on positive responses in the Main Questionnaire For agencies that had activities or planned activities related to facial recognition technology in the Main Questionnaire we asked that as appropriate the agencies complete an Attachment that covered the following areas 1 owned or accessed including tested facial recognition technology in fiscal year 2020 2 plans to own or access including testing facial recognition technology through fiscal year 2023 3 research and development conducted or supported in fiscal year 2020 4 any transactions the agency entered into with nonfederal entities for that entity to obtain facial recognition technology in fiscal year 2020 and 4GAO Face Recognition Technology FBI Should Better Ensure Privacy and Accuracy GAO-16-267 Washington D C May 16 2016 Facial Recognition CBP and TSA are Taking Steps to Implement Programs but CBP Should Address Privacy and System Performance Issues GAO-20-568 Washington D C Sept 2 2020 Facial Recognition Technology Commercial Uses Privacy Issues and Applicable Federal Law GAO-15-621 Washington D C July 30 2015 and Facial Recognition Technology Privacy and Accuracy Issues Related to Commercial Uses GAO-20-522 Washington D C July 13 2020 Page 33 GAO-21-526 Facial Recognition Technology Appendix I Objectives Scope and Methodology 5 any regulation of a nonfederal entity’s use of its own facial recognition technology 5 We used these timeframes because fiscal year 2020 was the most recent fiscal year for which information was available when we issued our questionnaire and through fiscal year 2023 for planned systems because agencies were most likely to have information covering this timeframe such as in strategic plans Each of the Attachments had detailed questions such as a description of the activity how the agency used the technology including the purpose s and obligations related to the activity For how agencies used facial recognition technology we asked whether the federal agency owned or accessed it directly e g logging into a system or via a third party e g asking another entity to run the search on its behalf For example we asked the name of the technology and the entity that owned it including how best to describe that entity e g federal state tribal or local For obligations we asked the agency to provide the amount obligated in fiscal year 2020 and whether it was disbursed If the agency could not provide an obligated amount we asked for a range or the reason why an agency did not know e g facial recognition was a small part of a larger biometric system and was not specifically tracked Finally we also asked if agencies had classified systems 6 After determining the areas of inquiry we conducted pretests with five agencies to test the questionnaire’s applicability to all agencies and a variety of facial recognition technology uses and revised the questionnaire based on those pretests We selected agencies for pretests based on information provided during initial meetings with agency officials to capture a variety of facial recognition technologies and purposes in order to test different parts of our questionnaire For example National Aeronautics and Space Administration officials told us they used facial recognition technology for research and development so we focused the pretest on the research and development questions with that agency On 5Regulated refers to using regulatory authority over a nonfederal entity to regulate that entity’s use of its own FRT For the purposes of our questionnaire we defined “regulated” as regulatory functions in which the agency engaged including but not limited to investigatory and inspections activities taking enforcement actions prescribing requirements or guidance conducting oversight and maintaining performance standards 6This report only discusses unclassified systems Page 34 GAO-21-526 Facial Recognition Technology Appendix I Objectives Scope and Methodology the other hand Department of Defense officials were able to pretest the majority of the questions so we met with them multiple times Once the design process was completed we administered the questionnaire by email to agency liaisons or their designees for an agency-level response about facial recognition technology use To do this we instructed the liaisons to provide the agency-level response in the Main Questionnaire We asked the liaisons to disseminate the Attachments to relevant and knowledgeable facial recognition technology subject matter experts at the agency component bureau or office levels which the liaisons would consolidate into the Main Questionnaire and return it and all the Attachments to us 7 We provided detailed instructions in writing and in the questionnaire itself We also followed up by email and phone when appropriate to ensure that the agency liaisons received the questionnaire and to ask if they had any questions about it Once the liaison determined their agency’s survey responses were complete we asked the liaison to total the Attachments and enter the total number of completed Attachments on the Main Questionnaire to ensure we received all of the expected agency Attachments in their survey responses When agencies submitted their survey responses we conducted an initial review for completeness inconsistencies or logical errors within the responses We asked agencies to re-submit or clarify responses if necessary Because we surveyed and obtained responses from all 24 agencies in the population defined by our scope the summary results describing this group are not subject to errors from sampling and nonresponse However the practical difficulties of conducting any survey may introduce other errors such as • Difficulties in how a particular question is interpreted by respondents For example some agencies were not sure which Attachment to fill out or completed one for an activity that did not meet our criteria so we followed up with them to determine which was the most appropriate in some cases Furthermore as part of our analysis of agency responses we determined that some information could have been included correctly in more than one Attachment In one case for 7For example the Departments of the Treasury and Defense did not consolidate their responses We informed them that we would take the responses provided by these agencies and consolidate them on behalf of the agency and provide them with an opportunity to review this information in the report Page 35 GAO-21-526 Facial Recognition Technology Appendix I Objectives Scope and Methodology agencies that reported using facial recognition technology such as eye tracking as part of research unrelated to facial recognition for other purposes we determined that those technologies would only be reported as research and development activities even if some agencies reported it as a system “used ” In another case for agencies that filled out the regulatory attachment we excluded responses that did not meet both of our criteria 1 that an agency had authority over a nonfederal entity’s use of facial recognition technology and 2 actually used that authority to regulate the nonfederal entity’s use of facial recognition technology • Sources of information that are available to respondents differ across agencies For example one agency expressed concern to us about their ability to provide a comprehensive response that included every accessed facial recognition technology because they did not track this information We asked that agencies provide what they could but to focus their efforts on facial recognition technology access that had a memorandum of agreement or understanding Furthermore we relied on the agency liaisons to provide Attachments to subject matter experts on facial recognition technology use within their agencies We provided suggestions to these liaisons of possible areas where facial recognition technology could be used such as security and information technology offices to ensure that the agency-level responses were comprehensive To help corroborate the information agencies provided in the questionnaire we conducted a search of government contracting information and reviewed information provided for prior reports Specifically we conducted a search of several terms such as “facial recognition ” “FRT ” and known vendor names in the Federal Procurement Data System Next Generation and Grants gov databases We used two analysts to independently review and determine if a result was related to FRT in our scope such as excluding non-human related results and other technologies We used this list to review each agency’s responses for completeness When we discovered discrepancies we followed up with the agency as appropriate to change their response or fill out new Attachments from the questionnaire Furthermore we reviewed responses from a survey of federal law enforcement use of facial recognition technology to determine if there Page 36 GAO-21-526 Facial Recognition Technology Appendix I Objectives Scope and Methodology were inconsistencies in responses to our survey 8 If there were we requested clarification of the agency-provided information such as system descriptions in interviews and multiple rounds of follow up i e information requests to those agencies For example if an agency filled out an Attachment but indicated “no” on the Main Questionnaire for the applicable question related to the Attachment we determined that was an incorrect response on the Main Questionnaire • How we processed and analyzed the responses we received can influence the accuracy of the survey results For example we consolidated some agency responses such as the Department of Defense We independently verified the consolidated information internally and presented it to those agencies prior to issuing the report We took steps in the development of the questionnaire such as pretesting data collection and data analysis including multiple rounds of follow up through interviews and information requests as noted above to minimize these potential errors and to help ensure the accuracy of the answers obtained Based on these quality assurance and control actions we determined that for the purposes of this report the information provided is an accurate and valid representation of the extent of facial recognition use across the 24 CFO Act agencies We conducted this performance audit from April 2020 through August 2021 in accordance with generally accepted government auditing standards Those standards require that we plan and perform the audit to obtain sufficient appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives We believe that the evidence obtained provides a reasonable basis for our findings and conclusions on our audit objectives 8For more information about the survey of federal law enforcement’s use of facial recognition technology see GAO Facial Recognition Technology Federal Law Enforcement Agencies Should Better Assess Privacy and Other Risks GAO-21-518 Washington D C June 3 2021 Page 37 GAO-21-526 Facial Recognition Technology Appendix II Summaries of Selected Federal Agencies’ Facial Recognition Technology Activities Appendix II Summaries of Selected Federal Agencies’ Facial Recognition Technology Activities This appendix provides summaries of those Chief Financial Officers Act agencies that reported facial recognition technology FRT activities in fiscal year 2020 and planned FRT systems through fiscal year 2023 We asked the following questions of each agency which correspond to the responses provided in each agency’s summary 1 “At any point in fiscal year 2020 did your agency use facial recognition technology for any of the following purposes By ‘use ’ we mean your agency 1 owned and or operated facial recognition technology for internal agency purposes 2 accessed another federal or nonfederal entity’s including local government or private company facial recognition technology under an agreement or arrangement as part of an agency program or activity or 3 tested facial recognition technology as part of a pilot proof of concept trial or evaluation for potential agency use ” 2 “Does your agency have plans to begin using facial recognition technology including upgrading a system to include facial recognition for internal agency use between fiscal year 2020 and fiscal year 2023 In other words your agency has not yet begun to use the facial recognition technology but it has taken steps to begin by fiscal year 2023 By ‘plans ’ we mean your agency has initiated a process to use facial recognition technology which could include an ongoing acquisition process a contract with a vendor or another agency a memorandum of understanding or a budget request Do not include hypothetical or exploratory conversations about potential uses of facial recognition technology within your agency ” 3 “In fiscal year 2020 did your agency conduct research and development R D involving facial recognition technologies This could include funding another entity to conduct R D on your agency’s behalf R D includes basic research applied research or experimental development technology readiness levels 1-6 R D could include developing facial recognition algorithms or evaluating existing algorithms ” 4 “In fiscal year 2020 did your agency enter into transactions to enable a nonfederal entity to obtain facial recognition technology for their own uses In other words the support financial or in-kind would enable nonfederal entities to develop purchase or use facial recognition technology for their own uses—not for your agency’s use By ‘transactions ’ we mean your agency awarded grants entered into contracts leases or cooperative agreements provided direct loans or Page 38 GAO-21-526 Facial Recognition Technology Appendix II Summaries of Selected Federal Agencies’ Facial Recognition Technology Activities loan guarantees or entered into any other transactions using other transactional authority ” 5 “In fiscal year 2020 did your agency engage in any regulatory functions over nonfederal entities that use facial recognition technology For our purposes ‘regulatory functions’ includes but is not limited to investigatory and inspections activities taking enforcement actions prescribing requirements or guidance conducting oversight and maintaining performance standards ” Specifically we provide summaries for the following 16 agencies • Department of Agriculture • Department of Commerce • Department of Defense • Department of Energy • Department of Health and Human Services • Department of Homeland Security • Department of the Interior • Department of Justice • Department of State • Department of Transportation • Department of the Treasury • Department of Veterans Affairs • General Services Administration • National Aeronautics and Space Administration • National Science Foundation • Social Security Administration The information in these summaries is from survey responses or requests for more information from the agencies We present details on FRT systems reported by agencies that own their FRT systems or access the FRT systems of other government entities including federal and nonfederal governments and commercial facial recognition service providers Agencies can have direct access to an FRT system such as by logging into the system or indirect access such as by requesting a Page 39 GAO-21-526 Facial Recognition Technology Appendix II Summaries of Selected Federal Agencies’ Facial Recognition Technology Activities state government i e a third party run a facial recognition search on behalf of the agency We also present obligations for FRT systems as applicable Each agency provided information on obligations in its survey response However we are presenting obligations data for informational purposes only because we did not corroborate it through other means such as document requests and some agencies reported facial recognition obligations as part of a larger biometric system We do not include information on obligations related to unlocking smartphones or tablets because agencies reported obligations of 1 none or free because the smartphones were included in the service contract 2 the cost of individual devices or 3 the total of all purchased smartphones 1 The facial recognition feature of the smartphones was included with the phones so there is no specific obligation The Department of Education the Department of Housing and Urban Development the Nuclear Regulatory Commission and the Small Business Administration do not have summaries because they reported they had no FRT activities in fiscal year 2020 and no plans to have FRT activities through fiscal year 2023 The Department of Labor the U S Agency for International Development the Environmental Protection Agency and the Office of Personnel Management do not have summaries because they reported that they only use facial recognition to unlock smartphones or tablets 1The 14 agencies that reported using facial recognition to unlock smartphones or tablets are the Departments of Agriculture Commerce Homeland Security Energy Justice Health and Human Services the Interior the Treasury Veterans Affairs the Environmental Protection Agency the National Aeronautics and Space Administration the National Science Foundation the Office of Personnel Management and the U S Agency for International Development Page 40 GAO-21-526 Facial Recognition Technology The U S Department of Agriculture USDA reported that it did not have facial recognition technology FRT activities in fiscal year 2020 However it reported plans to use a new FRT system through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 None reported FRT System Obligations in Fiscal Year 2020 None reported Research Development in Fiscal Year 2020 None reported Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 USDA reported it plans to use two other FRT systems through fiscal year 2023 both of them for physical security purposes and one of them also for domestic law enforcement purposes U S Department of Agriculture USDA Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status IDEMIA VisionPass is a facial recognition device that will be used to verify the identity of personnel for access within secure areas of a facility USDA plans to use the technology beginning in fiscal year 2022 The Avigilon Control Center facial recognition software will assist with accelerating response times by identifying individuals of interest based on secure watch lists USDA plans for the facial recognition software to be operational by fiscal year 2022 if funding is approved New federal FRT systems Agricultural Research Service IDEMIA VisionPass Office of Safety Security and Protection Avigilon Control Center Source GAO analysis of survey results GAO-21-526 Page 41 GAO-21-526 Facial Recognition Technology Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to USDA the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • • • • • • • • • • Agricultural Marketing Service Animal and Plant Health Inspection Service Economic Research Service Farm Service Agency Food and Nutrition Service Food Safety and Inspection Service Foreign Agricultural Service Forest Service National Agricultural Statistics Service National Institute of Food and Agriculture Natural Resources Conservation Service Office of the Chief Information Officer Office of the Chief Economist Page 42 • • • • • • • • • • • • Office of the General Counsel Office of the Inspector General Office of Partnerships Public Engagement Office of the Assistant Secretary for Civil Rights Office of Tribal Relations Office of Homeland Security Office of Operations Risk Management Agency Rural Development Rural Utilities Service Rural Housing Service Rural Business-Cooperative Service GAO-21-526 Facial Recognition Technology The Department of Commerce reported facial recognition technology FRT activities in fiscal year 2020 Specifically Commerce reported owning one FRT system and conducting FRT-related research and development R D in fiscal year 2020 Commerce also reported plans to use one other FRT system through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 Commerce reported it owned one FRT system in fiscal year 2020 for physical security purposes Department of Commerce Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Other Commerce and federal users Description of system use Commerce-owned FRT system Office of Chief Information Officer TYCO StoneLock infrared biometric facial recognition device This system uses FRT to control access to a secure data center by infrared facial matching as part of its three-factor authentication process Commerce reported that it did not continue to use the system after fiscal year 2020 • None Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 Commerce reported it did not obligate funds in fiscal year 2020 because the technology was purchased in 2016 Research Development in Fiscal Year 2020 Commerce reported conducting FRT-related R D specifically by the National Institute of Standards and Technology NIST to support federal agencies and commercial vendors in fiscal year 2020 as follows • • NIST supported the development of standards and methods in performance measurement image quality testing and evaluating technologies and interoperability for FRT NIST also runs the Facial Recognition Vendor Test program which evaluates the performance of various facial recognition algorithms against a host of performance metrics such as accuracy and speed in facial identification matching system performance across different demographics and more recently facial detection of individuals wearing masks due to COVID-19 Commerce reported that while there was no FRT-specific funding it obligated approximately $500 000 in fiscal year 2020 for biometrics research and standards work at NIST Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Page 43 GAO-21-526 Facial Recognition Technology Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 Commerce reported plans to use one other FRT system through fiscal year 2023 for physical security purposes Department of Commerce Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status This system uses infrared facial matching to control access to a secure data center as part of a three-factor authentication process Commerce is constructing a data center with this physical security system Commerce expects to move to the new data center by August 2021 New federal FRT system Office of the Chief Information Officer Continuity of Operations Data Center access control system Source GAO analysis of survey results GAO-21-526 Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to Commerce the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • • • • Bureau of Economic Analysis Bureau of Industry and Security U S Census Bureau Economic Development Administration Office of the Under Secretary for Economic Affairs International Trade Administration Minority Business Development Agency Page 44 • • • • • • National Oceanic and Atmospheric Administration National Technical Information Service National Telecommunications and Information Administration Office of the Inspector General Office of the Secretary U S Patent and Trademark Office GAO-21-526 Facial Recognition Technology The Department of Defense DOD reported facial recognition technology FRT activities in fiscal year 2020 Specifically DOD reported using nine FRT systems and conducting research and development R D DOD also reported plans to use three other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 DOD reported it used nine FRT systems in fiscal year 2020 including seven owned by the department and two systems owned by other entities that DOD accessed Of those nine FRT systems six are for domestic law enforcement three are for physical security five are for national security and defense and one is for identification card enrollment purposes Some of these systems are used for multiple purposes Department of Defense DOD Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Description of system use Other DOD and federal users DOD ABIS contains a database of militarycollected biometrics of foreign nationals including faces It is used to identify threat actors related to terrorism or counterintelligence as well as to research information about a person of interest or identify an individual for an investigative lead • The Defense Facial Comparison Tool is used to compare a law enforcement probe photo to a DOD captured photo to identify or verify DOD affiliated individuals • DOD U S Navy and U S Air Force RAPIDS is DOD’s enterprise tool for identification card issuance It uses facial detection to ensure the consistent size of a picture printed on the DOD identification card during identification card issuance • DOD all components Other agencies that issue these identification cards e g the Uniformed Services TacID Guard Dog is used to monitor camera feeds of individuals seeking access to DOD facilities for possible matches to a watchlist of potential threats • None The device uses FRT to control access to the Pentagon Force Protection Agency door by verifying the identity of authorized individuals • DOD U S Army Javelin is biometrics software installed on mobile devices The FRT system collects facial images from the devices and submits them to DOD’s ABIS to identify an individual in a criminal investigation This FRT system also performs immediate facial fingerprint and iris matching to individuals on watch lists • None DOD-owned FRT systems U S Army Department of Defense Automated Biometric Identification System DOD ABIS Defense Manpower Data Center Defense Facial Comparison Tool Defense Manpower Data Center Real-Time Automated Personnel Identification System RAPIDS U S Navy TacID Guard Dog Pentagon Force Protection Agency M C Dean access control device U S Air Force InCadence Ares Javelin Page 45 • • • • DOD multiple components Department of Homeland Security Department of Justice Department of State GAO-21-526 Facial Recognition Technology U S Navy Facial Automated Biometric Identification System FABIS Mobile FABIS Mobile is an FRT phone application that can hold a watchlist DOD personnel use it during public events such as airshows to take photos of individuals who act erratically or suspiciously to determine if they are a potential match to the watchlist If they are a match then other biometrics may be taken to verify the individual’s identity if they do not match DOD personnel send the photo for a manual comparison Accessed federal FRT systems Department of Justice Next Generation Identification Interstate Photo System NGI IPS Department of Homeland Security Automated Biometric Identification System IDENT • None DOD users NGI IPS can be used by DOD personnel to identify suspected terrorists or for counterintelligence For example if an individual is identified as a potential threat actor DOD may forward the request for a search through NGI IPS IDENT can be used to verify an individual’s identity determine whether an individual in two separate photos is the same person and compare a probe photo against images stored in IDENT to create a list of potential matches For example if an individual is identified by DOD as a potential threat actor DOD may forward the request for a search through IDENT Source GAO analysis of survey results GAO-21-526 Note DOD components that reported classified systems not described above include U S Central Command Defense Intelligence Agency U S European Command U S Indo-Pacific Command and U S Special Operations Command FRT System Obligations in Fiscal Year 2020 DOD reported obligating funds to operate its own FRT systems in fiscal year 2020 • • DOD reported obligating funds for four of the seven FRT systems it owned in fiscal year 2020 o DOD reported obligating about $3 6 million for DOD ABIS of which a portion is used for facial recognition about $100 000 for TacID Guard Dog $865 000 for the Defense Facial Comparison Tool and $80 000 for FABIS Mobile o DOD did not report obligating funds for Javelin or the facial detection portion of RAPIDS noting that the specific obligation for the FRT portion of that system is unknown because it was included as part of many changes to the RAPIDS system in fiscal year 2020 DOD also reported it obligated no funds for the M C Dean access control device because it was purchased prior to fiscal year 2020 DOD reported it did not obligate funds for the two FRT systems it accessed in fiscal year 2020 because federal partners provided them at no cost Research Development in Fiscal Year 2020 DOD reported conducting FRT-related R D to support its mission needs For example DOD reported researching new capabilities for RAPIDS which would support identity verification during online identification card renewal and personal identification number reset requests Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Page 46 GAO-21-526 Facial Recognition Technology Planned Use of FRT Systems through Fiscal Year 2023 DOD reported it plans to use three other FRT systems through fiscal year 2023 Specifically of those three FRT systems it plans to use two for physical security one for national security and defense and one for domestic law enforcement Of these one system will be used for multiple purposes Department of Defense DOD Recognition Technology FRT Systems Plans for Facial through Fiscal Year 2023 FRT system Description of system use Planned new use status DARPA plans to use this FRT to pull a visitor’s record automatically if they have previously been to their Visitor Welcome Center It will be used to expedite visitor processing while reducing personnel data entry task errors and duplication As of May 2020 DARPA is seeking a license to the algorithm through the U S Army The U S Air Force plans to collect facial images with mobile biometric devices including phones to compare against Clearview AI’s repository of facial images from open sources for matching individuals The U S Air Force plans to evaluate the product as part of an operational pilot in June 2020 The U S Army’s FRT pilot at Redstone Arsenal performs digitized photo matching in real time at automated access control points using registered Automated Installation Entry system volunteers It is intended to improve the access vetting process speed and reduce security risks The U S Army began the pilot in the second quarter of fiscal year 2021 New federal FRT system Defense Advanced Research Projects Agency DARPA Visitor Welcome System New access to commercial FRT system U S Air Force Clearview AI Evaluation of federal FRT system U S Army Automated Installation Entry Source GAO analysis of survey results GAO-21-526 Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to DOD the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • U S Transportation Command Department of Defense Chief Information Officer Page 47 • Department of Defense Office of the Inspector General GAO-21-526 Facial Recognition Technology The Department of Energy DOE reported facial recognition technology FRT activities in fiscal year 2020 Specifically DOE reported using one FRT system DOE reported it does not plan to use FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 DOE reported it owned one FRT system in fiscal year 2020 for physical security purposes Department of Energy DOE Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Description of system use Other DOE and federal users TacID Guard Dog performs facial matching and facial detection from live video DOE uses TacID Guard Dog to monitor entry and exit from controlled locations for personnel accountability and to log individuals' arrivals at an evacuation point • DOE-owned FRT system National Nuclear Security Administration TacID Guard Dog None Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 DOE reported it purchased TacID Guard Dog in December 2019 and obligated $150 000 for testing the FRT system during fiscal year 2020 Research Development in Fiscal Year 2020 None reported Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 None reported Page 48 GAO-21-526 Facial Recognition Technology Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to DOE the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • Office of the Chief Information Officer Office of the Inspector General Page 49 • Office of Science including National Laboratories GAO-21-526 Facial Recognition Technology The Department of Health and Human Services HHS reported facial recognition technology FRT activities in fiscal year 2020 Specifically HHS reported using four FRT systems and conducting research and development R D HHS reported plans to expand its use of other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 HHS reported it owned three FRT systems and accessed one commercial FRT system in fiscal year 2020 Of those four FRT systems one was for digital access one for physical security and three were for domestic law enforcement purposes One of these systems was used for multiple purposes Department of Health and Human Services HHS Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Description of system use Other HHS and federal users AnyVision allows real time facial matching from security camera videos and images CDC piloted using this system to supplement the manual review and memorization of watchlist faces by security guards at their facilities • None Centers for Medicare and Medicaid Services CMS Laptop unlocking This FRT system allows laptop users that have an appropriate camera to use FRT to unlock their laptop CMS conducted a small pilot of less than 20 participants using this feature on their laptops • None Office of the Inspector General OIG Griffeye Digital Investigate Pro Griffeye Digital Investigate Pro uses FRT to support automated analysis of digital video and images The OIG used the system in support of investigations involving child exploitation child sexual assault and trafficking to locate or identify victims • None HHS-owned FRT systems Centers for Disease Control and Prevention CDC AnyVision Accessed commercial FRT system Clearview AI HHS users Clearview AI is a facial image matching software system that operates as an internet search engine for faces using publicly available images such as from social media The OIG conducted an evaluation of the system in an attempt to identify unknown subjects of criminal investigation • OIG Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 HHS reported obligating funds for two of the three FRT systems it owned in fiscal year 2020 • HHS reported obligating $1 590 for Griffeye Digital Investigate Pro and $126 896 for AnyVision HHS reported it obligated no funds for the pilot using facial recognition to unlock laptops because the feature was included with the operating system Page 50 GAO-21-526 Facial Recognition Technology • HHS reported it obligated no funds for its access of Clearview AI because it was a free trial to evaluate the system Research Development in Fiscal Year 2020 HHS supported research involving facial analysis technology for purposes other than facial matching For example HHS’s National Institutes of Health awarded grants for research that use eye tracking as a tool for clinical research such as characterizing where children with and without autism spectrum disorder looked while following conversations in videos NIH does not specifically report dollar amounts tied to research eye tracking technology Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 HHS reported it plans to use one other FRT system through fiscal year 2023 for domestic law enforcement purposes Department of Health and Human Services HHS Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT System Description of system use Planned New Use Status Vintra will be used to search surveillance video for investigator-defined actions and objects such as directional movement vehicles or people The system does not maintain a database of known users but may be able to match a submitted image to an image on the surveillance video HHS has tested the system established security protocols and plans to begin operational usage of Vintra by the end of May 2021 New federal FRT system Office of the Inspector General Vintra Source GAO analysis of survey results GAO-21-526 Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to HHS the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • Administration for Children and Families Administration for Community Living Agency for Healthcare Research and Quality Agency for Toxic Substances and Disease Registry Page 51 • • • • Food and Drug Administration Health Resources and Services Administration Indian Health Service Substance Abuse and Mental Health Services Administration GAO-21-526 Facial Recognition Technology The Department of Homeland Security DHS reported facial recognition technology FRT activities in fiscal year 2020 Specifically DHS reported owning four FRT systems and accessing three federal FRT systems FRT systems in 17 states and five localities and two commercial FRT systems DHS also reported conducting research and development R D related to FRT entering into transactions with nonfederal entities for FRT and regulating nonfederal entities use of FRT DHS also reported it plans to use five other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 DHS owned four federal FRT systems and accessed multiple systems owned by other entities DHS used four FRT systems for domestic law enforcement six for border and transportation security and four for national security and defense purposes Some of these systems were used for multiple purposes DHS had access to at least 24 state local and commercial FRT systems for domestic law enforcement for border and transportation security and for national security and defense purposes Department of Homeland Security DHS Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Description of system use Other DHS and federal users U S Customs and Border Protection uses ATS’s FRT for the following populations 1 individuals seeking to enter or exit the United States whose names appear on a flight or vessel manifests or voluntary manifests submitted by bus or rail manifest “manifested travelers” 2 individuals applying for CBP programs facilitating travel to the United States and 3 subjects of interest who require additional research and analysis It matches photos for these three populations against a predetermined gallery of photos associated with derogatory information • TVS uses facial recognition to verify traveler identities upon arrival at or departure from ports of entry TVS compares a live photo of a traveler against a gallery of photos e g passport photos in DHS databases • DHS Transportation Security Administration IDENT offers facial recognition services to partners to verify an individual’s identity determine whether an individual in two separate photos is the same and compare a probe photo against images stored in IDENT for potential matches • DHS all components Department of Justice Department of Defense CAT-2 machines provide facial matching services to assist with identity verification of travelers by capturing live images at airport checkpoints AutoCat is an electronic gate version of CAT-2 • DHS-owned FRT systems U S Customs and Border Protection Automated Targeting System ATS U S Customs and Border Protection Traveler Verification Service TVS Office of Biometric Identity Management Automated Biometric Identification System IDENT Transportation Security Administration Self-Service Version of Credential Authentication Technology with Camera CAT-2 and AutoCata Page 52 • • • Department of Justice Department of State DHS Science and Technology Directorate GAO-21-526 Facial Recognition Technology Accessed federal FRT systems Department of Defense Automated Biometric Identification System DOD ABIS Department of Justice Next Generation Identification Interstate Photo System NGI IPS Department of State Integrated Biometric System IBS DHS users DHS uses DOD ABIS’s facial recognition to identify foreign nationals connected to a national security investigation or to a known terrorist organization • U S Immigration and Customs Enforcement DHS uses NGI IPS to identify individuals of interest Specifically U S Customs and Border Protection submits probe photos via ATS to IDENT and are forwarded to NGI IPS NGI IPS returns a list of potential matching photos • Office of Biometric Identity Management U S Customs and Border Protection DHS uses IBS to identify visa applicants for travel documents and individuals involved in identity theft and benefit fraud investigations DHS employees have direct and indirect access to IBS—performing facial recognition searches in IBS while others submit photos for matching from apprehensions and bookings among others • Accessed state and local FRT systems 15 states Alabama Arizona Delaware Indiana Kansas Kentucky Michigan Mississippi Missouri Nebraska North Dakota South Dakota Utah Virginia and West Virginia • • • U S Citizenship and Immigration Services U S Customs and Border Protection U S Immigration and Customs Enforcement DHS users U S Customs and Border Protection U S Immigration and Customs Enforcement DHS’s Homeland Security Information Network HSIN is a system for trusted sharing of Sensitive But Unclassified information between federal state local territorial tribal international and private sector partners HSIN contains a mechanism to request third party facial recognition searches through the listed state and local entities such as fusion centers • Michigan Law Enforcement Information Network MLEIN DHS uses MLEIN to identify individuals involved in a crime DHS submits U S Border Patrol apprehension photos and others to the Statewide Network of Agency Photos • U S Customs and Border Protection New York State Intelligence Center Photo Imaging Mugshot System PIMS DHS uses PIMS to identify unknown individuals of interest in state and federal cases including individuals who have been deported and reentered the country DHS has direct access to PIMS through Border Patrol Agents working at the New York State Intelligence Center • U S Customs and Border Protection Ohio Law Enforcement Gateway OHLEG DHS uses OHLEG to identify individuals involved in a crime DHS submits photos such as U S Border Patrol apprehension photos • U S Customs and Border Protection Pinellas County Face Analysis Comparison and Examination System FACES DHS uses FACES to identify unknown individuals to support operations criminal investigations and administrative cases DHS submits photos to the Pinellas County Sheriff’s Office • U S Customs and Border Protection 4 localities Chicago Greater Cincinnati Delaware Valley Southeast Florida Accessed commercial FRT systems Clearview AI Vigilant Solutions • DHS users DHS uses Clearview AI to identify unknown individuals of interest in state federal and international cases such as child exploitation cases DHS may submit photos e g surveillance photos or receive requests to match photos against the Clearview AI database • DHS uses Vigilant Solutions to identify individuals involved in a crime and submits photos for matching • • • U S Customs and Border Protection U S Immigration and Customs Enforcement U S Secret Service U S Customs and Border Protection Source GAO analysis of survey results GAO-21-526 Note DHS also reported using classified systems a This system was only tested and not deployed Page 53 GAO-21-526 Facial Recognition Technology FRT System Obligations in Fiscal Year 2020 DHS reported obligating funds to operate its own FRT systems or access other ones including state and commercial systems in fiscal year 2020 • • • DHS reported obligating funds for three of the four FRT systems it owns in fiscal year 2020 Specifically DHS reported obligating $7 96 million for the annual maintenance cost of the license for IDENT $61 5 million for TVS and $2 5 million for CAT-2 DHS reported it obligated no funds for ATS because ATS is funded through the license for IDENT DHS reported obligating funds for two FRT systems it accessed in fiscal year 2020 Specifically DHS reported obligating about $23 67 million for HSIN and $31 592 to fund approximately seven licenses under its contract for access to the Michigan State Police’s MLEIN For the remaining federal state and local FRT systems it accessed o DHS reported obligating no funds to access three federal FRT systems—DOD ABIS NGI IPS and IBS—because those systems are funded by DOD DOJ and State respectively o DHS reported obligating no funds for state and local FRT systems because they were accessed at no cost through respective state and local owners DHS reported obligating funds to access one of two commercial FRT systems in fiscal year 2020 Specifically DHS’ Immigration and Customs Enforcement reported obligating $214 000 to purchase Clearview AI licenses while Secret Service has not yet made a decision to purchase services after conducting a pilot in April 2019 U S Customs and Border Protection accessed Clearview AI at no cost through an agent stationed at the New York State Intelligence Center Finally DHS reported it has limited access to Vigilant Solutions at no cost Research Development in Fiscal Year 2020 DHS reported conducting R D for testing on one FRT system and additional research to improve face detection analysis and matching capabilities • • • • The Transportation Security Administration reported obligating over $2 17 million for independent testing and planning for CAT-2 The tests will validate facial matching performance through a series of pilots of the Transportation Security Administration’s specific use case at checkpoints The Science and Technology Directorate reported obligating $1 2 million to sponsor Biometric Technology Rallies which are ongoing industry events with challenges to develop innovative solutions for biometric collection and matching including facial recognition and the reliability of collecting and matching information on travelers wearing masks during COVID-19 The Science and Technology Directorate and Office of Biometric Identity Management reported obligating about $553 000 for National Science Foundation’s Center for Identification Technology Research university grants Specifically S T sponsors R D and awards grants to university partners that focus on work related to face detection analysis and matching OBIM sponsors R D with a portion of work including fingerprint matching presentation attack detection and matching of biometrics from juveniles The Science and Technology Directorate reported obligating $300 000 to sponsor R D on face detection analysis and matching through its interagency agreement with the National Institute of Standards and Technology Transactions with Nonfederal Entities in Fiscal Year 2020 DHS reported that it entered into agreements with foreign governments and private entities and awarded grants to local governments and other entities • The Transportation Security Administration reported it had a Cooperative Research and Development Agreement with a technology company and an air carrier focused on ways to make mobile digital identity credentials e g mobile digital driver’s licenses interoperable with Transportation Security Administration’s checkpoint security systems Page 54 GAO-21-526 Facial Recognition Technology • • • The Science and Technology Directorate entered into project arrangements with two foreign governments—Australia and the United Kingdom—related to the assessment of face recognition software DHS awarded a contract to the Lehigh County Pennsylvania District Attorney’s Office to enhance the U S Immigration and Customs Enforcement Homeland Security Investigation’s future access to the Gang Intelligence Application DHS reported spending about $200 000 on the application in fiscal year 2020 The Federal Emergency Management Agency reported it obligated about $1 25 million across several grants that included facial recognition such as equipment for emergency responders and state and local law enforcement Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 DHS reported regulated a nonfederal entity’s use of FRT Specifically the Transportation Security Administration issued security program amendments to Delta Air Lines to permit the use of facial identification technology to identify passengers checking baggage for air transportation Planned Use of FRT Systems through Fiscal Year 2023 DHS reported it plans to use five other FRT systems through fiscal year 2023 Of those five FRT systems three will be used for border and transportation security two will be used for national security and defense and two will be used for domestic law enforcement purposes Some of these systems will be used for multiple purposes Department of Homeland Security DHS Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status U S Immigration and Customs Enforcement plans to develop facial recognition access to its data in an automated criminal justice information system U S Immigration and Customs Enforcement is working with the Lehigh County District Attorney’s Office to complete access by May 2021 The Transportation Security Administration plans to initiate a new pilot of U S Customs and Border Protection’s TVS to evaluate use of biometric technology including facial recognition to automate identity verification at checkpoints and modernize screening The Transportation Security Administration began a pilot at the Detroit Metropolitan Wayne County Airport in March 2021 with U S Customs and Border Protection New access to local FRT system Lehigh County Pennsylvania District Attorney’s Office Gang Intelligence Application Evaluation of federal FRT system U S Customs and Border Protection Traveler Verification Service TVS Page 55 GAO-21-526 Facial Recognition Technology Department of Homeland Security DHS Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status Transportation Security Administration The Transportation Security Administration plans to test this technology to automate identity verification at checkpoints and modernize screening of travelers The Transportation Security Administration began demonstrating CAT-2 at Ronald Regan Washington National Airport in August 2020 In March 2021 it started field site testing of CAT-2 at additional airport checkpoints to identify evaluate and mitigate system performance issues across diverse operational environments and passenger demographics Data collected during field tests will be used for qualitative and quantitative analysis by the Science and Technology Directorate It is also testing and monitoring AutoCAT at the TSA Systems Integration Facility in preparation for future field pilots HART will replace IDENT’s capabilities initially to include FRT and additional capabilities will be added in subsequent years The Office of Biometric Identity Management plans to replace IDENT with HART by December 2021 Credential Authentication Technology with Camera System CAT-2 and AutoCAT Upgrade of federal FRT system Office of Biometric Identity Management Homeland Advanced Recognition Technology HART Source GAO analysis of survey results GAO-21-526 Note We included DHS’s plans to pilot an additional federal FRT system the details of which are sensitive in the reported number of planned use of FRT systems Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to DHS the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • U S Coast Guard Cybersecurity and Infrastructure Security Agency Federal Law Enforcement Training Center Page 56 • • • Countering Weapons of Mass Destruction Office of the Inspector General Office of Operations Coordination GAO-21-526 Facial Recognition Technology The Department of the Interior reported facial recognition technology FRT activities in fiscal year 2020 Specifically Interior reported accessing two FRT systems in fiscal year 2020 and planning to use two FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 Interior reported it accessed one locally owned FRT system and one commercial FRT system in fiscal year 2020 for domestic law enforcement Department of the Interior Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Other Interior and federal users Description of system use Accessed Local FRT system National Capital Region Facial Recognition Investigative Leads System NCRFRILS Interior users NCRFRILS is an FRT system that contains copies of information including photos from participating law enforcement agencies Interior requested a third party search of NCRFRILS to compare an image obtained from twitter as an example against the photo database to generate investigate leads Accessed commercial FRT system Clearview AI • Interior U S Park Police Interior users Clearview AI is an FRT system that can identify an individual through facial matching by comparing a photo against its facial image database Interior uses Clearview AI to verify the identity of an individual involved in a crime and research information on a person of interest Interior may submit photos e g surveillance photos for matching against the Clearview AI’s repository of facial images from open sources U S Park Police reported it stopped using Clearview AI as of June 2020 • Interior U S Park Police Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 Interior reported it obligated no funding to access NCRFRILS in fiscal year 2020 because Interior specifically the U S Park Police accessed this system through a third party—the Maryland National Capital Park Police In addition Interior reported it obligated no funding to access Clearview AI in fiscal year 2020 because it was a free trial of Clearview AI Research Development in Fiscal Year 2020 None reported Page 57 GAO-21-526 Facial Recognition Technology Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 Interior reported it plans new access to a local FRT system and a commercial FRT system through fiscal year 2023 for domestic law enforcement purposes Department of the Interior Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status NCRFRILS is an FRT system that contains copies of information including photos from participating law enforcement agencies Interior requested a third party search of NCRFRILS to compare an image obtained from twitter as an example against the photo database to generate investigate leads In April 2021 U S Park Police officials told us their plans to establish a memorandum of understanding for direct access to NCRFRILS Clearview AI is an FRT system that can identify an individual through facial matching by comparing a photo against its facial image database Interior will use Clearview AI to verify the identity of an individual involved in a crime and research information on a person of interest Interior may submit photos e g surveillance photos for matching against the Clearview AI repository of facial images from open sources Interior reported its U S Fish and Wildlife Service began using a trial version of Clearview AI in May 2020 and purchased an annual subscription in June 2020 New access to local FRT system National Capital Region Facial Recognition Investigative Leads System NCRFRILS New access to commercial FRT system Clearview AI Source GAO analysis of survey results GAO-21-526 Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to Interior the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • • • • • Bureau of Indian Affairs Bureau of Indian Education Bureau of Land Management Bureau of Ocean Energy Management Bureau of Reclamation Bureau of Safety and Environmental Enforcement Bureau of Trust Funds Administration Office of the Assistant Secretary for Policy Management and Budget Page 58 • • • • • Office of the Inspector General Office of the Secretary Office of Surface Mining Reclamation and Enforcement Office of the Solicitor U S Geological Survey GAO-21-526 Facial Recognition Technology The Department of Justice DOJ reported facial recognition technology FRT activities in fiscal year 2020 Specifically DOJ reported using federal state local and commercial FRT systems conducting research and development R D related to FRT and entering into transactions with nonfederal entities for FRT DOJ also reported plans to expand its use of FRT through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 DOJ reported it used 11 federal and commercial FRT systems in fiscal year 2020 in addition to a number of state and local systems Of those 11 FRT systems DOJ used eight for domestic law enforcement one for physical security two for national security and defense one for video management and two for educational purposes Some of these systems were used for multiple purposes All of the state and local systems were used for domestic law enforcement and some national security and defense purposes FRT system Department of Justice DOJ Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 Description of system use Other DOJ and federal users Horus is being tested to determine if it can be used to improve the accuracy and efficiency of one-to-one comparison processes by serving as an aid to examiners It is also used in educational settings to demonstrate how FRT works • None NGI IPS uses facial matching from an unknown image of interest to law enforcement against mugshots in the photo database to generate possible matches for investigators • • Department of Homeland Security Department of Defense Similar to Horus this system is being tested to determine if it can be used to improve the accuracy and efficiency of one-to-one comparison processes by serving as an aid to examiners It is also used in educational settings to demonstrate how FRT works • None This system uses FRT to control access by identifying and verifying individuals entering onsite secure network operations centers at federal prisons Federal Bureau of Prisons officials noted the system will not be in use after fiscal year 2022 • None At the request of U S law enforcement officials including state local tribal and federal authorities the U S National Central Bureau can send photos such as missing persons and suspects in investigative cases for comparison against the holdings in IFRS IFRS contains facial images received from more than 160 countries and member countries can provide facial images for matching These results are returned to the requesting country and also to the country that provided the images • Any U S law enforcement agency including federal via the U S National Central Bureau DOJ-owned FRT systems Federal Bureau of Investigation Horus Federal Bureau of Investigation Next Generation Identification Interstate Photo System NGI IPS Federal Bureau of Investigation RankOne Federal Bureau of Prisons Facial Recognition Access Control System U S National Central Bureau USNCB International Criminal Police Organization INTERPOL Facial Recognition System IFRS Page 59 GAO-21-526 Facial Recognition Technology U S Marshals Service Axon Facial Detection Axon Facial Detection is used to review footage from body-worn cameras for faces i e face detection that a human then selects deselects for redaction across multiple video frames Accessed federal FRT systems • None DOJ users Department of Defense Automated Biometric Identification System DOD ABIS DOD ABIS is owned and operated by the Department of Defense and is used to identify foreign nationals connected to a national security investigation such as a known terrorist organization or suspected of criminal activity DOJ can request an image search on DOD ABIS through FBI’s NGI IPS • Federal Bureau of Investigation Department of State Integrated Biometric System IBS IBS is owned and operated by the Department of State and is used to identify visa applicants for travel documents and individuals involved in identity theft and benefit fraud investigations Specifically FBI has direct access to perform facial recognition searches within the Department of State Consular Consolidated Database’s visa holdings and FBI’s Facial Analysis Comparison and Evaluation FACE Services has indirect access to passport photos FACE Services must request a Department of State passport officer perform facial recognition searches of passport photos on the FBI’s behalf • Federal Bureau of Investigation Accessed state and local FRT systems DOJ users 21 states Alabama Arizona Arkansas Colorado Delaware Idaho Illinois Indiana Iowa Kentucky Maryland Michigan Nebraska New Mexico North Carolina North Dakota Pennsylvania South Carolina Tennessee Texas and Utah The FBI’s FACE Services is an internal provider of facial recognition searches and requests for FBI investigations FACE Services examiners have direct or third party access through memorandums of understandings with 21 different states’ FRT systems FACE Services also has direct and indirect i e by request access to various holdings within the Department of State’s IBS and indirect access to DOD ABIS • Federal Bureau of Investigation 15 states Alabama Arizona Delaware Indiana Kansas Kentucky Michigan Mississippi Missouri Nebraska North Dakota South Dakota Utah Virginia and West Virginia DHS’s Homeland Security Information Network HSIN is a system for trusted sharing of Sensitive But Unclassified information between federal state local territorial tribal international and private sector partners HSIN contains a mechanism to request third party facial recognition searches through the listed state and local entities such as fusion centers • U S Marshals Service DOJ requested third-party searches of these systems separately by the state or local owner in order to identify fugitives individuals involved in a crime or people using fraudulent identification i e aliases among other persons of interest • U S Marshals Service Authorized DOJ employees can directly search NCRFRILS for matches against photos from participating local law enforcement agencies For example an examiner reviewed video evidence from a retail store’s surveillance cameras with facial recognition software comparing the still image from the video with regional booking photos • Bureau of Alcohol Tobacco Firearms and Explosives U S Marshals Service 4 localities Chicago Greater Cincinnati Delaware Valley Southeast Florida Other states Arkansas Arizona Ohio Pennsylvania and Tennessee Other localities Pinellas County Florida and San Diego California National Capital Region Facial Recognition Investigative Leads System NCRFRILS Page 60 • GAO-21-526 Facial Recognition Technology Accessed commercial FRT systems Clearview AI DOJ users DOJ uses Clearview AI to identify unknown individuals of interest in state federal and international cases such as child exploitation DOJ may submit photos e g surveillance photos or receive requests to match photos against the Clearview AI database • • • • Vigilant Solutions DOJ uses Vigilant Solutions to identify individuals involved in a crime For example an examiner reviewed photos from social media e g Facebook regarding a string of convenience store robberies and compared them with arrest photos using facial recognition software • Bureau of Alcohol Tobacco Firearms and Explosives Drug Enforcement Administration Federal Bureau of Investigation U S Marshals Service Bureau of Alcohol Tobacco Firearms and Explosives Source GAO analysis of survey results GAO-21-526 Note We included an additional FRT system owned by DOJ the details of which are sensitive in the reported number of FRT systems used DOJ also reported using classified systems FRT System Obligations in Fiscal Year 2020 DOJ reported obligating funds to operate its own FRT systems and to access commercial ones in fiscal year 2020 • • • DOJ reported obligating funds for one of the nine FRT systems it owns in fiscal year 2020 Specifically DOJ reported obligating $17 million for its biometrics contract which included the facial recognition algorithm used for NGI IPS o DOJ reported it obligated no funds for the following systems  Facial Recognition Access Control System because it was purchased in 2014  Axon Facial Detection because the vendor upgraded the capabilities at no cost to a system already in procurement and  Horus and RankOne because they were developed by DOJ or by a partner agency that provided copies to DOJ at no cost DOJ reported obligating funds for none of the federal state and local FRT systems it accessed in fiscal year 2020 because respective partners provided them at no cost For example DOJ reported that access to the INTERPOL Face Recognition System is also at no cost because it is a dues paying member of the INTERPOL organization DOJ reported obligating funds to access one of two commercial FRT systems in fiscal year 2020 Specifically DOJ obligated $9 000 for some Clearview AI licenses and all access to Vigilant Solutions at no cost through federal state and local partner agencies Research Development in Fiscal Year 2020 DOJ reported conducting and supporting R D to support its mission needs Specifically the FBI reported obligating $1 56 million for several types of FRT research as follows • • An interagency agreement with the National Institute of Standards and Technology for 1 testing and evaluation of current industry face image quality tools and identification of best practices 2 a facial recognition algorithm benchmark for testing overall accuracy and continued analysis of the effect demographics has on accuracy and 3 benchmark testing technology to detect face image manipulation such as a deepfake with the goal of discerning industry capabilities Applied research at the West Virginia University Research Corporation into 1 the relationship between skin tone and false match rates in facial recognition algorithms and 2 assessing the capabilities and limitations of current synthetic face detection such as deepfakes and developing synthetic face detection software prototypes Page 61 GAO-21-526 Facial Recognition Technology • FBI’s Operational Technology Division is conducting applied research on facial matching For example the Horus and RankOne systems were tested for the potential benefits of combining FRT systems with trained forensic examiners for verification Transactions with Nonfederal Entities in Fiscal Year 2020 DOJ reported that it awarded an $836 000 grant to the Police Foundation for the development of techniques to automate analysis of body worn camera audio and video data of police and community interactions In particular these techniques could 1 allow an evaluation of officers’ adherence to principles of procedural justice and 2 validate the ratings generated by the automated process using a randomized control trial comparing software ratings of videos to evaluations performed by human raters under conditions of high and low procedural justice Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 DOJ reported it plans to use two other FRT systems through fiscal year 2023 Of these one will be used for border and transportation security and one will be used for physical security purposes Department of Justice DOJ Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status DEA plans to use the facial recognition component of a security video management system to control access monitor and surveil its headquarters facility This system is not operational as of June 2021 because DEA’s headquarters facility is being renovated DEA plans to begin using the FRT features after completing a privacy assessment USMS plans to develop software to perform touchless prisoner identity verification such as during prisoner transport searching for a match against booking and prisoner photos within the JPATS mobile application for transportation security purposes USMS planned to begin using the JPATS Mobile Application in mid-to-late 2021 but has been delayed because the bid was protested New Federal FRT systems Drug Enforcement Administration DEA Security video management system U S Marshals Service USMS Justice Prisoner and Alien Transportation System JPATS Mobile Application Source GAO analysis of survey results GAO-21-526 Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to DOJ the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • • • U S Attorneys Office on Violence Against Women Community Oriented Policing Services Office of Information Policy Foreign Claims Settlement Commission DOJ Divisions Antitrust Civil Civil Rights Criminal Environment Natural Resources Page 62 • • • • • Justice Management Tax and National Security Community Relations Service Office of the Solicitor General Office of Professional Responsibility Office of the Inspector General Office of the Pardon Attorney GAO-21-526 Facial Recognition Technology • • • U S Parole Commission Executive Office for Immigration Review Executive Office for Organized Crime Drug Enforcement Task Forces Page 63 • • Executive Office for U S Trustees Professional Responsibility Advisory Office GAO-21-526 Facial Recognition Technology The Department of State reported facial recognition technology FRT activities in fiscal year 2020 Specifically State reported owning one FRT system and accessing one FRT system in fiscal year 2020 State also reported conducting FRT-related research and development R D and entering into transactions with nonfederal entities for FRT State reported it plans to use one other FRT system through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 State reported it owned one FRT system and accessed one other FRT system in fiscal year 2020 for border and transportation security and national security and defense purposes Department of State Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Other State and federal users Description of system use State-owned FRT system Bureau of Consular Affairs Integrated Biometric System IBS State uses IBS to verify an applicant’s identity or determine whether an individual has applied for a visa State uses FRT to search visa and passport photos contained in IBS for matches with individuals who previously applied for travel documents or may be involved in visa fraud Potential matches from IBS are sent to the Kentucky Consular Center and National Visa Center for review before a final determination is made by consular officers at posts or passport specialists at passport agencies and centers Accessed federal FRT system Department of Defense Automated Biometric Identification System DOD ABIS • • • Department of State Bureau of Diplomatic Security Department of Justice Department of Homeland Security State users State uses DOD ABIS to verify the authenticity of travel documents State submits multimodal biometric files e g fingerprints and face and iris scans to DOD ABIS through a U S Special Operations Command portal or the Department of Justice’s Next Generation Identification Interstate Photo System • Bureau of Diplomatic Security Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 In fiscal year 2020 State reported it obligated over $8 5 million for IBS In addition State reported it obligated no funds to access DOD ABIS because DOD funded the FRT system Research Development in Fiscal Year 2020 State reported conducting the following FRT-related R D to enhance facial analysis and facial matching in fiscal year 2020 as follows Page 64 GAO-21-526 Facial Recognition Technology • • State’s Bureau of Consular Affairs reported conducting R D to enhance facial analysis including morphing detection e g making a synthetic face image using two faces to allow both individuals to use the image for identification purposes and the impact of aging on the accuracy of facial recognition algorithms such as for children’s passports State reported this R D will contribute to International Organization for Standardization and International Civil Aviation Organization image standards for travel documents State reported obligating over $1 7 million for these projects in fiscal year 2020 State’s Bureau of Counterterrorism reported conducting R D to enhance the facial matching of screening technology used in the Personal Identification Secure Comparison and Evaluation System PISCES border management system The system matches images of individuals against passport images and a repository of images of suspicious individuals State reported obligating approximately $3 million for a larger effort to improve program technology including the enhanced screening technology Transactions with Nonfederal Entities in Fiscal Year 2020 State reported it entered into the following transactions to assist foreign governments with purchasing and using previously donated FRT equipment • • State’s Bureau of International Narcotics and Law Enforcement reported entering into an interagency agreement and contract with Mexico’s National Migration Institute Specifically State assisted with building the institute’s capacity to collect store and share biometric data on third-country nationals with donated FRT equipment In addition State reported it funded technical advisors to train Mexican government employees to use the equipment State reported obligating approximately $2 6 million for the technical assistance and training State’s Bureau of International Narcotics and Law Enforcement also reported it obligated approximately $333 000 to purchase 10 workstations for the Guatemalan Immigration Institute The institute plans to install the workstations at three locations—the La Aurora International Airport the Guatemalan Immigration Institute Detention Facility and the Valle Nuevo port of entry on the El Salvador-Guatemala border Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 State reported it plans to use one other FRT system through fiscal year 2023 for border and transportation security Department of State Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status The FRT system matches individuals against passport images and a repository of known or suspected terrorists using the system State reported plans to incorporate enhanced screening technologies in PISCES to ensure foreign partners under the Terrorist Interdiction Program are able to protect themselves from attempts by terrorists to enter transit or depart their country State plans to pilot the new FRT system in the summer of 2021 If the testing is successful State will deploy the software to operational locations New federal FRT system Bureau of Counterterrorism Personal Identification Secure Comparison and Evaluation System PISCES Source GAO analysis of survey results GAO-21-526 Page 65 GAO-21-526 Facial Recognition Technology Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to State the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • Office of the Secretary of State Office of the Deputy Secretary Office of the Under Secretary for Arms Control and International Security Page 66 • • • Office of the Under Secretary for Economic Growth Energy and Environment Office of the Under Secretary for Political Affairs Office of the Under Secretary for Public Diplomacy and Public Affairs GAO-21-526 Facial Recognition Technology The Department of Transportation DOT reported conducting facial recognition technology FRT related research and development R D activities in fiscal year 2020 Specifically DOT used FRT to conduct human factors research in a variety of transportation-related areas DOT reported it did not have any other FRT activities in fiscal year 2020 and does not plan to use FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 None reported FRT System Obligations in Fiscal Year 2020 None reported Research Development in Fiscal Year 2020 DOT reported the following R D projects in fiscal year 2020 that used FRT specifically facial detection and analysis to conduct human factors research such as • • • • The Federal Motor Carrier Safety Administration conducted human factors research involving commercial motor vehicle drivers’ behavior and safety performance For example an onboard monitoring system analyzes a driver’s face to identify fatigue and distraction while driving The FRT has eye glance analysis which measures visual attention or inattention rates drowsiness and measures the percent of eye closure as an indicator of fatigue The Federal Highway Administration used FRT at its lab which has a simulator and test vehicle to conduct human factors research They use the results to incorporate highway driver needs into roadway design construction repair and improvement In addition it obligated $150 580 through an interagency agreement to the National Science Foundation’s Big Data Hubs program to support research into better ways to protect the privacy of drivers who participated in a mandated naturalistic driver study that recorded their incabin behavior This research is intended to improve other researchers’ access to the driver data helping them use it to improve traffic safety It also obligated $300 000 to Oak Ridge National Laboratory which included facial detection research The Federal Railroad Administration reported operating the Cab Technology Integration Lab—a full-sized locomotive cab simulator—that uses an eye tracking device and software to determine what a train engineer is looking at on a display screen or when scanning the environment outside the cab This data helps researchers determine what objects in the environment capture the engineer’s attention while driving a train The Federal Aviation Administration reported conducting human factors research that uses eye tracking technology to observe how air traffic controllers scan their instruments to determine eye motion workload and the time spent looking at each object This research is intended to ensure that systems that include human operators and maintainers perform as effectively and safely as possible DOT reported it could not provide specific obligations for individual R D projects that used FRT because the programs do not track obligations at that level Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Page 67 GAO-21-526 Facial Recognition Technology Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 None reported Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to DOT the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • Office of the Secretary of Transportation Office of Inspector General National Highway Traffic Safety Administration Federal Transit Administration Page 68 • • • Pipeline and Hazardous Materials Safety Administration Great Lakes Saint Lawrence Seaway Development Corporation Maritime Administration GAO-21-526 Facial Recognition Technology The Department of the Treasury reported facial recognition technology FRT activities in fiscal year 2020 Specifically Treasury reported accessing two FRT systems Treasury also reported plans to use three other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 Treasury reported it accessed two FRT systems in fiscal year 2020 one for digital access and one for domestic law enforcement purposes Department of the Treasury Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Other Treasury and federal users Description of system use Accessed federal FRT system General Services Administration login gov Treasury users This login gov pilot used FRT that compared two photos to verify the identity of an individual accessing the website or application Login gov takes a picture of the individual and their photo identification to determine a match using FRT Accessed commercial FRT system Vendor facial recognition search services • Internal Revenue Service IRS Treasury users A third-party vendor performed facial recognition searches on behalf of the IRS for domestic law enforcement purposes Additional details on the search are sensitive • IRS Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 Treasury reported it obligated no funding to access the two FRT systems in fiscal year 2020 Specifically GSA funded its login gov FRT system and a vendor provided a demonstration of their FRT system at no additional cost Research Development in Fiscal Year 2020 None reported Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Page 69 GAO-21-526 Facial Recognition Technology Planned Use of FRT Systems through Fiscal Year 2023 Treasury reported it plans to use three FRT systems through fiscal year 2023 for domestic law enforcement purposes Department of the Treasury Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status Facebox identifies facial images of similar persons linked to multiple investigations through an online storage locker that contains information including photos from other investigations The FRT system notifies the investigators of potential matches TIGTA purchased Facebox in June 2020 and plans to begin use of the FRT system in December 2020 The MOUs will allow IRS agents direct access to other federal biometric databases including facial recognition The additional access will assist IRS agents with criminal investigations For example an IRS agent may submit a surveillance photo to the National Forensic Laboratory which will conduct a facial recognition search against one of the FRT systems The IRS’s National Forensic Laboratory is in the process of establishing MOUs with the Departments of Defense Homeland Security and Justice As of April 2021 the IRS was determining whether separate MOUs are needed A third-party vendor performed facial recognition searches on behalf of the IRS for domestic law enforcement purposes Additional details on the search are sensitive IRS plans to conduct an additional pilot pending additional funding New federal FRT system U S Treasury Inspector General for Tax Administration TIGTA Facebox New access to federal FRT system Internal Revenue Service IRS Memoranda of Understanding MOUs to access federal FRT systems Evaluation of commercial FRT system IRS Vendor facial recognition search services Source GAO analysis of survey results GAO-21-526 Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to Treasury the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • Alcohol and Tobacco Tax and Trade Bureau Bureau of Engraving and Printing Bureau of the Fiscal Service Financial Crimes Enforcement Network Page 70 • • • • Office of Inspector General Office of the Comptroller of the Currency U S Mint Treasury Departmental Offices GAO-21-526 Facial Recognition Technology The Department of Veterans Affairs VA reported facial recognition technology FRT activities in fiscal year 2020 Specifically VA reported conducting FRT-related research and development R D and entering into transactions with nonfederal entities for FRT VA reported plans to use two other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 None reported FRT System Obligations in Fiscal Year 2020 None reported Research Development in Fiscal Year 2020 VA reported using FRT as a tool to conduct research for purposes other than facial matching Specifically VA reported using eye tracking to conduct clinical research for treating post-traumatic stress disorder The eye tracking system evaluates pupil response to evaluate impairment VA reported obligating $22 840 for the equipment Transactions with Nonfederal Entities in Fiscal Year 2020 VA purchased two types of eye tracking equipment for veterans • • VA reported it purchased prosthetics that enable veterans with speech impairment or loss to communicate using a computer or tablet device VA reported purchasing 42 devices for $416 284 VA reported it purchased a prosthetic device that enabled one veteran to use eye gaze to manipulate a laptop computer for $18 860 Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 VA reported it plans to use two FRT systems through fiscal year 2023 for physical security and domestic law enforcement purposes Page 71 GAO-21-526 Facial Recognition Technology Department of Veterans Affairs VA Plans for Facial Recognition Technology FRT Systems through Fiscal Year 2023 FRT system Description of system use Planned new use status The FRT system can sort through video quickly locating a specific person such as a missing patient or to input a photo from a police bulletin e g ‘Be On the Lookout’ to alert officers when that person enters the VA medical center campus VA reported that they plan to obligate funding for a project design to install Motorola Avigilon by fiscal year 2022 The web-based FRT can detect and follow selected moving objects such as a previously disruptive individual or to track missing patients and will alert officers when on the VA medical center campus VA purchased FRT software for 12 cameras in 2019 and is working with the contractor to ensure the system is operational by the middle of fiscal year 2022 New federal FRT systems VA Police Service Chicago IL Motorola Avigilon VA Police Service West Palm Beach FL Veritone ai Ware Source GAO analysis of survey results GAO-21-526 Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to VA the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • Veterans Benefits Administration Page 72 • National Cemetery Administration GAO-21-526 Facial Recognition Technology The General Services Administration GSA reported facial recognition technology FRT activities in fiscal year 2020 Specifically GSA reported owning one FRT system GSA reported it does not plan to use other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 GSA reported it owned one FRT system in fiscal year 2020 for digital access purposes General Services Administration GSA Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Description of system use Other GSA and federal users Login gov conducted a pilot with Department of the Treasury employees to test FRT services that compared two photos to verify the identity of an individual accessing the website or application Login gov takes a picture of the individual and their photo identification to determine a match • GSA-owned federal FRT system Technology Transformation Service login gov Department of the Treasury Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 GSA reported it obligated over $90 000 for login gov in fiscal year 2020 Research Development in Fiscal Year 2020 None reported Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 None reported Page 73 GAO-21-526 Facial Recognition Technology Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to GSA the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • National Services Public Building Service Regional Offices Staff Offices Page 74 • Independent Offices Office of Inspector General GAO-21-526 Facial Recognition Technology The National Aeronautics and Space Administration NASA reported facial recognition technology FRT activities in fiscal year 2020 Specifically NASA reported testing one FRT system and conducting FRT-related research and development R D NASA reported it does not plan to use other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 NASA reported it owned one FRT prototype system in fiscal year 2020 Specifically NASA tested this system for issuing badges National Aeronautics and Space Administration NASA Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 FRT system Other NASA and federal users Description of system use NASA-owned FRT system Johnson Space Center Forgotten badge prototype The forgotten badge prototype confirms an employee's identity by comparing a current camera image of the employee with a photo on file NASA reported that it would not continue work on the prototype due to cost • None Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 NASA reported it did not obligate funds for its forgotten badge prototype in fiscal year 2020 because funding for the prototype was obligated in fiscal year 2019 Research Development in Fiscal Year 2020 NASA reported using FRT as a tool to conduct human factors research Specifically the Langley Research Center conducted a series of controlled research experiments from 2013 to 2020 to understand the cognitive states of aircraft and space flight crew These experiments included the use of eye tracking and facial muscle tracking devices to understand cognitive states e g surprised focused and other human factors during simulations NASA reported it obligated $60 000 for the R D experiments in fiscal year 2020 Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 None reported Page 75 GAO-21-526 Facial Recognition Technology Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to NASA the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • Administrator Staff Offices Mission Directorates Mission Support Directorate Page 76 • • Office of the Administrator Office of the Inspector General GAO-21-526 Facial Recognition Technology The National Science Foundation NSF reported facial recognition technology FRT activities in fiscal year 2020 Specifically NSF reported it supported FRT-related research and development R D in fiscal year 2020 NSF also reported it does not plan to use other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 None reported FRT System Obligations in Fiscal Year 2020 None reported Research Development in Fiscal Year 2020 NSF reported it supported FRT-related R D conducted by external organizations in fiscal year 2020 Specifically NSF’s Directorate for Computer and Information Science and Engineering awarded three grants to universities and others to conduct research on FRT and related areas For example • • • The University of Chicago conducted research using machine learning for facial recognition The Columbus State University conducted research on facial privacy The University of Nebraska conducted research using eye tracking to optimize programmer productivity NSF reported obligating $1 696 146 for these grants in fiscal year 2020 but this amount includes areas of research other than FRT Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 None reported Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to NSF the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • Chief Information Officer Directorate for Biological Sciences Page 77 • • Directorate for Education and Human Resources Directorate for Engineering GAO-21-526 Facial Recognition Technology • • • • • Directorate for Geosciences Directorate for Mathematical and Physical Sciences Directorate for Social Behavioral and Economic Sciences National Science Board Office of Inspector General Office of Budget Finance and Award Management Page 78 • • • • • • • Office of the Director Office of Diversity and Inclusion Office of the General Counsel Office of Information Resource Management Office of Integrative Activities Office of International Science and Engineering Office of Legislative and Public Affairs GAO-21-526 Facial Recognition Technology The Social Security Administration SSA reported facial recognition technology FRT activities in fiscal year 2020 Specifically SSA reported accessing one FRT system in fiscal year 2020 SSA reported it does not plan to use other FRT systems through fiscal year 2023 Use of FRT Systems in Fiscal Year 2020 SSA reported it accessed one FRT system in fiscal year 2020 for digital access purposes FRT system Social Security Administration SSA Facial Recognition Technology FRT Systems Used in Fiscal Year 2020 Description of system use Other SSA and federal users Accessed commercial FRT system Acuant FaceID SSA users This pilot of Acuant FaceID was conducted using agency employees to verify the identity of employees within the test group for access to SSA’s public online services by remotely confirming that the facial image on an identity document e g state-issued IDs matches the facial image of the applicant • Office of Digital Transformation Source GAO analysis of survey results GAO-21-526 FRT System Obligations in Fiscal Year 2020 SSA reported that it could not provide the specific amount that was obligated for FRT because this FRT was included as part of a larger task order Instead SSA reported that it obligated $809 179 as part of a larger contract to support digital identity enhancements including the pilot Research Development in Fiscal Year 2020 None reported Transactions with Nonfederal Entities in Fiscal Year 2020 None reported Regulation of Nonfederal Entities’ Use of FRT in Fiscal Year 2020 None reported Planned Use of FRT Systems through Fiscal Year 2023 None reported Page 79 GAO-21-526 Facial Recognition Technology Components with No FRT-Related Activities in Fiscal Year 2020 or Planned Uses through Fiscal Year 2023 According to SSA the following components reported no FRT-related activities in fiscal year 2020 or planned uses through fiscal year 2023 listed above • • • • • • Office of Analytics Review and Oversight Office of Budget Finance and Management Office of the Chief Actuary Office of the Commissioner Office of Communications Office of General Counsel Page 80 • • • • • • Office of the Inspector General Office of Hearings Operations Office of Human Resources Office of Legislation and Congressional Affairs Office of Operations Office of Retirement and Disability Policy GAO-21-526 Facial Recognition Technology Appendix III Comments from the U S Agency for International Development Appendix III Comments from the U S Agency for International Development Page 81 GAO-21-526 Facial Recognition Technology Appendix III Comments from the U S Agency for International Development Page 82 GAO-21-526 Facial Recognition Technology Appendix IV Comments from the Social Security Administration Appendix IV Comments from the Social Security Administration Page 83 GAO-21-526 Facial Recognition Technology Appendix V GAO Contacts and Staff Acknowledgments GAO Contacts Candice N Wright at 202 512-6888 or wrightc@gao gov Gretta L Goodwin at 202 512-8777 or goodwing@gao gov Staff Acknowledgments In addition to the contact named above Adam Hoffman Assistant Director and Richard Hung Assistant Director Katrina Pekar-Carpenter Analyst-in-Charge Kelsey Burdick Jehan Chase Nirmal Chaudhary Caitlin Cusati Khaki LaRiviere Sarah Prokop and Carl Ramirez made key contributions to this report Also contributing were Christina Bixby Cheron Brooks April Gillens Sig Janoska-Bedi Tom Lombardi Robert Rivas and Benjamin Shouse 104231 Page 84 GAO-21-526 Facial Recognition Technology GAO’s Mission The Government Accountability Office the audit evaluation and investigative arm of Congress exists to support Congress in meeting its constitutional responsibilities and to help improve the performance and accountability of the federal government for the American people GAO examines the use of public funds evaluates federal programs and policies and provides analyses recommendations and other assistance to help Congress make informed oversight policy and funding decisions GAO’s commitment to good government is reflected in its core values of accountability integrity and reliability Obtaining Copies of GAO Reports and Testimony The fastest and easiest way to obtain copies of GAO documents at no cost is through our website Each weekday afternoon GAO posts on its website newly released reports testimony and correspondence You can also subscribe to GAO’s email updates to receive notification of newly posted products Order by Phone The price of each GAO publication reflects GAO’s actual cost of production and distribution and depends on the number of pages in the publication and whether the publication is printed in color or black and white Pricing and ordering information is posted on GAO’s website https www gao gov ordering htm Place orders by calling 202 512-6000 toll free 866 801-7077 or TDD 202 512-2537 Orders may be paid for using American Express Discover Card MasterCard Visa check or money order Call for additional information Connect with GAO Connect with GAO on Facebook Flickr Twitter and YouTube Subscribe to our RSS Feeds or Email Updates Listen to our Podcasts Visit GAO on the web at https www gao gov To Report Fraud Waste and Abuse in Federal Programs Contact FraudNet Website https www gao gov about what-gao-does fraudnet Automated answering system 800 424-5454 or 202 512-7700 Congressional Relations A Nicole Clowers Managing Director ClowersA@gao gov 202 512-4400 U S Government Accountability Office 441 G Street NW Room 7125 Washington DC 20548 Public Affairs Chuck Young Managing Director youngc1@gao gov 202 512-4800 U S Government Accountability Office 441 G Street NW Room 7149 Washington DC 20548 Strategic Planning and External Liaison Stephen J Sanford Managing Director spel@gao gov 202 512-4707 U S Government Accountability Office 441 G Street NW Room 7814 Washington DC 20548 Please Print on Recycled Paper