Unfortunately, this practice isn’t hypothetical—U.S. Customs and Border Protection (CBP) is currently field testing a new facial recognition system. This technology is intended to improve surveillance capabilities at border crossings. Spanning 152 days of testimony, the trial itself unfolded from late 2021 through early 2022. Its stated purpose was to photograph people coming into the United States by car. Dave Maass, director of investigations at the Electronic Frontier Foundation, received a copy of the document after filing a public records request. Those results from this initial test were originally reported by The Intercept.
Customer and border protection CBP’s use of facial recognition technology is built on a one-to-one matching process. This new system matches an individual’s photo to their travel credentials. In reality, for instance, at the Anzalduas border crossing, cameras successfully photographed all occupants of vehicles about 76% of the time. Of those photographed, 81 percent passed the validation requirements for matching their facial images to identification documents.
CBP’s attempts to improve surveillance tactics isn’t a new trend. The agency has long advocated for better tracking infrastructures. This commitment has continued across administrations, a recent welcome and obvious signal that the Foundation is intent on broadening its reach. This is a troubling assumption to make given the practical limitations and execution of such surveillance methods.
“The primary risk of CBP’s face recognition system is the system failing to recognize that someone matches their own documents,” Maass stated. This very real concern points to a deeper potential for harm in the system. These typos can result in false identifications and delays at land border crossings. Though we know a bit about what’s driving these error rates. Are the improvements due to the increasing quality of cameras, or are they a result of matching algorithm efficiency improvements?
CBP is leading a deeper effort, which they are calling ImmigrationOS. Its purpose is to obtain what it calls “near real-time visibility” on people who are self-deporting from the United States. The ImmigrationOS platform aims to clarify the numbers behind self-deportations, supplementing biometrically confirmed entries in travelers’ crossing histories. This broad approach represents a cumulative attempt to surveil, monitor, and control people inside the borders of the United States.
“CBP’s surveillance strategy carries over from administration to administration. It always falls short. It always has vendor issues and contracting issues and waste issues and abuse issues,” Maass remarked. This important finding clearly conveys that even with the best technology, the same longstanding issues continue to ail CBP’s surveillance operations.
In addition, as Maass reminded us, we need to be critical about the demographic effects that these systems can have. “We don’t know what racial disparities, gender disparities, etc., come up with these systems,” he noted, emphasizing that without proper oversight, these technologies may perpetuate existing biases.
In rhetoric, CBP purports that its facial recognition system improves passenger images and identifies 100% of vehicle passengers. The sheer breadth of this surveillance scheme makes it deeply alarming. Without robust transparency and accountability provisions, these efforts risk violating civil liberties and privacy rights, critics say.
The obstacles encountered by CBP’s facial recognition program are indicative of the larger issues in law enforcement with this emerging technology. CBP’s approach contrasts with the one-to-many facial recognition systems that police have deployed. It places an emphasis on biometric verification by matching people, one-to-one, with their own identification documents.
Leave a Reply