Clear spars over TSA report after security incident follows enrollment error

Biometric security has debuted in a Korean airport and there are plans to install it in Kazakhstan. But a private-industry implementation in the U.S. by biometric verification firm Clear is experiencing a problem.

How problematic the Clear deployment is is the source of heated debate, but if nothing else, the argument highlights the one fact few in the industry discuss. AI algorithms, even when collaborating with humans, are imperfect.

U.S. business news publisher Bloomberg last week reported that a Clear subscriber a year ago was almost able to board a flight departing Washington National Airport under a fake identity. The suspect allegedly was caught only when a government inspection spotted live ammunition in the person’s possession.

Bloomberg’s account, which it says includes previously unknown details, indicates that the U.S. Transportation Security Administration investigated the incident and came down hard on Clear.

The government found that some travelers were able to enroll in Clear using Clear-captured photos that contained no recognizable faces. The algorithm reportedly issued an alert on the images, but employees allegedly overrode the flags.

Travel trade publisher One Mile at a Time posted four images reportedly from investigation.

The investigation found that 49,000 Clear subscriptions were flagged by AI because images of individuals did not match.

Clear executives almost immediately and emphatically issued a statement saying Bloomberg’s article was far off base. They emphasized the time the company spent with the reporters and expressed incredulity that their efforts were not rewarded with vindication.

The executives say “the TSA has reverified 4.7 million IDs without citing a single issue” in the last six months. They are also upset that the government gave reporters the four example photos, although no one can be identified in them.

What is more, according to the executives, Clear uses fingerprints and irises to verify subscribers the day they travel, not face biometrics.

Meanwhile, Incheon Airport, in South Korea, is using pre-registered facial information for departures, according to reporting by The Korea Herald. The new system has now reached availability at 16 gates, and makes it possible to bypass verification completely, including at the boarding gate.

The service is limited to six airlines right now, according to The Herald. Travelers can use an Apple or Google app to complete an application or use an airport kiosk to register their face template for use in the service. The facial recognition system is expected to be used throughout the airport by April 2025.

The facial recognition software being used is from Alchera and covers check-in, bag-drop, security screening and boarding.

Kazakhstan is behind Korea in this regard. The Central Asia nation has created a commission to consider the implementation of airport biometrics. Commissioners are still addressing questions like how to ensure fluid interactions between the national government and its relevant agencies, according to reporting by The Astana Times.

They reportedly are nearing “effective completion” of their International Civil Aviation Organization biometrics audit. The government intends to move into biometrics for travel with a project at the Nursultan Nazarbayev International Airport.

Article Topics

airport biometrics  |  biometrics  |  contactless biometrics  |  facial recognition  |  identity verification  |  Kazakhstan  |  passenger processing  |  South Korea  |  United States