CBP Facial Recognition Program Has Gathered 50 Million Face Photos, Identified Fewer Than 300 Imposters
from the more-buck-than-bang-at-this-point dept
The CBP and DHS have released their annual report [PDF] covering trade and travel. It touts the agencies’ successes in these areas but raises some questions about the use of facial recognition tech to make the nation safer.
Dave Gershgorn, writing for One Zero, points out the system the DHS and CBP claim is essential to national security isn’t doing much to secure the nation. And it’s not for a lack of input data.
U.S. Customs and Border Protection scanned more than 23 million people with facial recognition technology at airports, seaports, and pedestrian crossings in 2020, the agency recently revealed in its annual report on trade and travel.
The agency scanned four million more people than in 2019. The report indicates that the system caught no imposters traveling through airports last year and fewer than 100 new pedestrian imposters.
There are two ways of looking at this — neither of which justify the DHS’s aggressive expansion of biometric collection at airports, border crossings, and any other place travelers might be found.
This low number of imposters recognized could indicate the system is flawed — incapable of accurately doing the job it’s supposed to do: recognize faces. This is a problem inherent to every facial recognition algorithm being used anywhere. The tech tends to be most accurate when presented with white male faces. Everyone else is subjected to higher risk of being misidentified, which can lead to false arrests or being prevented from traveling.
But it could also indicate imposters aren’t all that common — certainly not common enough to justify the erection of a mass surveillance system that gives travelers the option of giving up their biometric info or not traveling.
The CBP only began tracking and publishing stats in 2018, trailing its biometric rollout by a couple of years. Since 2018, the government has caught less than 300 imposters. Over that same time period, it performed 50 million face scans. That’s a pretty terrible return on investment. But that’s not how the CBP portrays this in its report:
Biometrics have proven an effective tool to combat the use of stolen and fraudulent travel and identity documents. Since the program’s inception, in 2018, CBP officers at U.S. airports have successfully intercepted seven impostors who were denied admission to the United States and identified 285 imposters on arrival in the land pedestrian environment.
I guess it all depends on what your definition of “effective” is. The CBP’s expansive definition suggests literally any number above zero justifies the cost of the tech and its attendant surveillance creep.
Of course, if the real purpose of the program is to create a massive database of facial photos and personal information, it’s been a tremendous success. And if that’s all the CBP and DHS expect from it, it will continue to perform this task with minimal interruption.
But travelers and taxpayers should expect more from the DHS and its agencies. Spending millions to deal with a minor problem by deploying tech that remains unproven shouldn’t be considered acceptable. Neither is the alternative: a system that rarely recognizes imposters, allowing government agencies to assume it’s less of a problem than it might actually be.