A U.S. government watchdog agency called for stronger face recognition protocols for federal agencies.

What’s new: An audit of federal agencies by the Government Accountability Office (GAO) found that, while many employ face recognition, they may not know where it came from, how it’s being used, or the hazards involved. The auditors recommended that agencies using commercial systems develop protocols for appropriate use.

What they found: Twenty agencies that employ law-enforcement officers reported using face recognition.

  • Of these, 11 used systems developed by private companies including Clearview AI and Vigilant. The others either developed their own or used systems developed by another agency. One of the most popular is the Department of Homeland Security’s Automated Biometric Identification System, which contains data on 835 million individuals.
  • Several agencies did not seem to know who built some of the systems they use.
  • Six agencies used the technology to investigate people involved in protests against police brutality. Three used it to look into perpetrators of the January 6 attack on the U.S. Capitol.
  • Only one agency that reported using a privately developed face recognition system — Immigrations and Customs Enforcement — had implemented oversight protocols such as requiring that employees report each use.

Behind the news: Face recognition is increasingly controversial in the U.S. Lawmakers recently introduced legislation that would freeze government use of the technology. At least 20 U.S. cities and several states have passed laws that restrict the technology.

Why it matters: Face recognition has clear potential to infringe on privacy. Moreover, it has a spotty record of identifying minorities, which has led to false arrests. The finding that many federal agencies are taking a cavalier approach raises troubling questions about privacy and fairness.

We’re thinking: The GAO audit of face recognition systems is a step forward. While regulators, ethicists, technologists, and businesses sort out appropriate standards, a moratorium on law enforcement use of face recognition would be sensible, so we can position the technology for socially beneficial uses while guarding against detrimental ones.


Subscribe to The Batch

Stay updated with weekly AI News and Insights delivered to your inbox