[Vendor-funded] Think tank: Consider worries about border AI but don’t touch the brake
U.S. border-control officials need to stop listening so intently to traveling Americans worried about the government’s growing use of facial recognition and other biometric tools. They have money to expand systems, and their departments need to get to it.
A global-policy think tank’s new report about biometrics at the border says biometric systems have increased port security and operational streamlining, and is critical of public worries about their accuracy and effects on a democratic society.
The border biometrics report, about balancing security, convenience and civil liberties, is published by the Atlantic Council and paid for by government contractor (and biometrics vendor) SAIC.
Its authors are unabashed supporters of AI for the tedious process of trying to find and stop everyone border forces need to, from those traveling under false identities to those involved in terrorist networks.
They back up some of their enthusiasm with nuggets that are hard to debate.
For example, the Customs and Border Protection agency uses NEC Corp‘s NEC-3 facial recognition algorithm. The National Institute of Standards and Technology in 2019 judged that code to be “on many measures, the most accurate we have evaluated.”
And there is already a technological foundation for expansions.
The current Traveler Verification System launched in federal fiscal year 2016. That same year, Congress budgeted Customs up to $100 million annually for 10 years to collect biometric data from foreign nationals leaving the country.
In 2020, CBP completed its first biometrics installation, at Los Angeles International Airport, for identifying people entering and exiting the United States. That program, Simplified Arrival, continues to expand across the nation.
Periodic, complete statistics on these operations have not been published.
But after listing three very real worries some Americans express about biometrics at the border — a growing surveillance state, data security, and system mistakes — the authors largely dismiss them.
Congressional and new, dedicated government departments can provide oversight, keeping border biometrics safe from executive branch abuse, according to the report.
It waves off concern about data breaches with a fallacious straw man defense. Security threats are not unique to CBP, so if theft or misuse is a threat to all government, which it is, then no corner of the federal government should collect any private data because it cannot be safeguarded.
Finally, accuracy and bias problems reportedly are so rare with the NEC-3 that it seems unworthy of measuring wrong “no-match” ID determinations. The agency has processed 130 million travelers since 2017, and erroneous determinations “have not arisen as a major issue.”
But then the authors take the wind out of their big-number argument by boasting “more than two thousand imposters” (emphasis by the authors) have been caught using port biometrics.
That is not a big number compared with 1.3 million to 2.6 million, which is how many people over roughly four years who the face biometrics algorithm has failed to match, based on the DHS-calculated accuracy rate of 98 to 99 percent.
The report offers four recommendations for moving biometrics forward on the borders.
First, CBP should continue to improve its facial comparison capabilities. And early work on a Simplified Arrival program for vehicles should also continue.
Homeland Security, CBP’s parent department, should create organizations to address unresolved questions about data retention and sharing.
Third, department leaders should go full-bore, spending every appropriated dollar, including those allocated as part of the Infrastructure Investment and Jobs Act. And Congress should appropriate more funds just to be safe.
Last, the Biden administration should really listen to public comments regarding operations before approving programs that are “close to the form” originally outlined in November 2020.