President Donald Trump’s controversial travel ban called for, among other things, the speedy completion of a “biometric entry-exit tracking system” for all travelers to the United States.
If this sounds familiar, it’s because the idea has been debated in Washington for more than a decade. The implementation of such a system was one of the recommendations from the sprawling document known as the 9/11 Report, published 13 years ago by the National Commission on Terrorist Attacks Upon the United States.
In fact, members of Congress mandated the creation of an enhanced entry-exit database before the attacks of 2001, as part of immigration reform in 1996. After the September 11 attacks, Congress set a 2006 deadline for the implementation the system, and specified that agencies government-wide—not just “scattered units at Homeland Security and the State Department”— should be able to access it. When the federal government missed that deadline, Congress issued a new target for 2009.
Eight years later, it still hasn’t happened. There are several obstacles to creating the kind of system that officials in Washington have demanded. The accuracy of biometric identification systems and the cost of building such a system in the first place—plus government-wide computer upgrades that would be required to support its use—are all major considerations. Plus, airlines have so far refused to follow government rules that say they should collect and process biometric data from passengers leaving the United States.
In the meantime, the technological landscape has changed dramatically. Advances in facial recognition software and long-range iris scanning—plus the mass adoption of smartphones—mean that a biometric entry-exit system could be far more expansive in 2017 than when such a system was first proposed.
Biometric systems of the past collected fingerprints. Today’s systems can be built to recognize individual faces, even voiceprints. In the not-too-distant future, they will be able to identify someone by subtle behavioral cues: how they swipe their fingers across a touch screen, for example. With the advent of long-range iris scanning and the ubiquity of surveillance cameras, systems are also likely to become more passive—meaning, they’ll do the work of identifying people without requiring much (or any) active engagement from the person being identified.
Already, most foreign nationals are required to have biometric data—fingerprints, passport photos—collected or checked when they arrive in the United States. Airline carriers and commercial ships collect information about people leaving the country—lists of passengers, for example—and share those manifests with U.S. Customs and Border Protection, which maintains its own database.
The idea behind a new biometric entry-exist system is to add layers of authenticating data—fingerprinting, iris scanning, facial recognition—to verify the identity of a person who is leaving the country, matching records against what was collected upon entry. Keeping track of foreigners who are coming and going, the thinking goes, could prevent a terrorist attack from being carried out in the United States by non-citizens overstaying their visas. The “large majority of jihadist terrorists in the United States,” however, have been American citizens or legal residents, according to a terrorism-tracking project by the think tank New America. “Every jihadist who conducted a lethal attack inside the United States since 9/11 was a citizen or legal resident,” according to New America’s research.
There’s no reliable data reflecting how many people have entered the country and overstayed their visas, anyway. The Department of Homeland Security had a backlog of nearly 2 million “unmatched arrival records,” each indicating a foreign national who had entered the country but not yet left, according to a 2013 Government Accountability Office report. Yet Homeland Security records don’t take into account approved immigration status changes—like cases in which a person is granted a deferred departure.
The Office of Biometric Identity Management, a division of Homeland Security, says it has already stopped “thousands of people who were ineligible to enter the United States.” At the same time, state-level identity bureaus are increasingly turning to biometrics for verifying that people are who they say—when they seek a driver’s license, for example.
The identity-services firm MorphoTrust now partners with most states—35 of them by last count—on biometric data-collection systems. “We’ve got a great model,” said Bob Eckel, the MorphoTrust CEO, referring to how his firm’s work with the states could shape a federal biometric entry-exit system. And technology has finally reached a point, he says, where a federal system could be cost efficient.
So far, even with federal funding and ongoing pilot programs, money has been a major sticking point. “Despite the call by some lawmakers for an exit system, airports and the airline industry have balked because it would cost airlines $3 billion, according to a 2013 Homeland Security estimate,” The New York Times reported last year. “The department issued regulations in 2008 requiring airports to collect biometric exit information, but carriers have largely ignored the regulation, and there have been no sanctions.”
The reliability of technological systems is, along with cost, the other substantial hurdle. What’s technically possible doesn’t always align with what’s practical.
MorphoTrust already makes systems that use facial recognition software and machines that can collect fingerprints with the wave of a hand. Long-range iris scanning is still prohibitively expensive, Eckel says—but it’s coming.
In the meantime, it would be easy to build a system that could use a network of cameras to verify travelers’ identities in real time as they move through an airport, Eckel told me. Such security systems could link up with mobile-apps so that a person could take a selfie while waiting in line at Customs to speed up the process. (The security-selfie is an idea MorphoTrust has floated for verifying a person’s identity in other sensitive transactions, such as credit card purchases.)
Eckel makes it sound almost effortless, but there are still “major physical infrastructure, logistical, and operational hurdles to collect an individual’s biographic and biometric data upon departure,” according to a 2015 Homeland Security report.
And despite significant technological advances in recent years, the possibility of misidentifications remains a serious issue. Even in the best facial-recognition systems, accuracy plummets as datasets grow.
Identical twins are just one of many puzzling challenges for recognizing algorithms, a MorphoTrust vice president told me last fall. Machines also have trouble telling lookalikes apart—doppelgangers who aren’t genetically related—and systems can be easily tripped up by differences in lighting or angle. (Machine systems can even get stumped by someone making a goofy face. That’s why, one computer scientist recently told me, she wouldn’t trust a machine to tell the difference between Tom Hanks and Bill Murray in a widely circulated photo that confused plenty of humans, too.)
Civil liberties advocates like the ACLU say the collection of biometric data poses “an extraordinary threat to privacy.” The possibility of inaccuracies made by a biometric exit-entry system pose a graver threat still. “Overreactions can impose high costs,” the 9/11 Commission wrote in its 2004 report, “on individuals, our economy, and our beliefs about justice.”
An equally pressing concern, though, is how human judgment is applied to such systems—in cases of false positives and legitimate threats. “Four of the 9/11 attackers were pulled into secondary border inspection, but then admitted,” the 9/11 report says. “More than half of the 19 hijackers were flagged by the Federal Aviation Administration’s profiling system when they arrived for their flights, but the consequence was that bags, not people, were checked.”