Facial recognition systems will be released in late August, according to one US Court of Auditors report (Government Accountability Office, GAO) “Increasingly” used: most commonly in cybersecurity, domestic law enforcement, and physical security. Eighteen of the 24 federal agencies surveyed are currently using at least one type of facial recognition, and many are using more than one system. Most are federally owned systems, but six are from commercial providers like Clearview AI, Vigilant Solutions, and Acuant FaceID.
Ten ministries are also planning to expand their use of facial recognition by 2023 and implement 17 different facial recognition systems: the agriculture, economic, defense, homeland security, health, justice and finance ministries, as well as the interior and foreign ministries and the ministry for veterans affairs. Thirteen of these systems will be federal, two will be owned by local law enforcement, and two will be using Clearview AI.
The results are coming after a year of public protests by advocates of privacy and civil liberties against police and government use of the technology. Face recognition has been found to be less accurate in people with darker skin, women, and younger and older people. Another GAO report recently released also cited a lack of oversight from federal law enforcement agencies using the technology.
Diverse use of face recognition
Many agencies are already using or planning to use facial recognition to protect sensitive data and technology and physical locations, or are focused on judicial and military matters. For example, the Inspector General’s Office began assisting Vintra in May by searching surveillance videos for “directed movement, vehicles, or people.” The US Marshals Service is developing a contactless identification system for prisoners to facilitate the booking and transportation of prisoners. The U.S. Immigration and Customs Service is working with the Lehigh County District Attorney’s Office in Pennsylvania to expand access to their existing facial recognition system through integration with an eGang intelligence application.
The Department of Agriculture plans to use facial recognition systems to monitor live video surveillance for people on watch lists if funding is approved. Both the US Air Force, as part of the Department of Defense, and the Fish and Wildlife Service in the Department of the Interior are working on projects with Clearview AI and want to expand its use. Clearview AI is controversial because its matching algorithm is based on a database with over three billion public images based, which were skimmed from the Internet. Other facial recognition systems used by law enforcement agencies are based on much smaller databases, often developed by the government: this includes, for example, a database of photos of people accused of a crime.
Ten agencies also conduct research and development in this area, including the Justice, Defense, Homeland Security and State Ministries. The agencies’ goals are diverse, but many investigate the well-documented distortions of many facial recognition systems. For example, the Department of Justice is researching the relationship between skin tone and false identification rates of facial recognition algorithms. Others looked at how to make such systems more accurate, even when scanning people wearing masks.
“Growing dependence of the federal government”
The report also demonstrated extensive cross-agency coordination and sharing of facial recognition systems and information. Many federal agencies reported that they obtained their facial recognition systems from state and local governments. For example, the Department of Homeland Security stated that its information network “contains a mechanism for soliciting third-party facial recognition searches through the listed state and local entities, such as fusion centers.” In fusion centers, various police, intelligence and other agencies work together to combat threats.