America is using AI to transform airport body scanning
High-speed body scanners powered by advanced artificial intelligence are to be trialled across US airports and subway stations.
According to The Guardian, documents filed by Evolv Technology with the US Federal Communications Commission (FCC) show plans to test the new systems at Denver International airport, Los Angeles’ Union Station metro and Union Station in Washington DC.
Evolv’s scanners use millimetre-wave radio frequencies to scan a person’s body. This setup is currently used by the full-body scanners found in airports, but whereas those devices slow down the screening process, the new scanners can finish their work in a fraction of a second. This means individuals can simply walk through the gate without having to stop.
Evolv claims its system can get through 800 people in an hour, without the need for people to remove smartphones or items of clothing.
Alongside the scanners, a guard will work with a tablet that will either show an “all-clear” symbol, or a picture of the individual with suspicious areas on their body highlighted. Crucially, the human guard is unable to see exactly what the scanner is looking for. Instead, an AI system works through data, picked up using radar beams from solid-state micro-antennas in the scanner. Working through this information, the AI calculates the likelihood of the person having guns or bombs on their body.
The scanner also takes a photograph of each person that passes through it, which is intended to allow facial recognition of individuals.
“Transportation is a very soft and attractive target,” Alex Wiggins, the executive in charge of security for the Los Angeles County Metropolitan Transportation Authority, told The Guardian. “Given the recent large-scale attacks at transit facilities in Europe, we need to see if there is technology that can screen [a] large number of peoples and focus in on weapons and explosives.”
Evolv, which includes Microsoft founder Bill Gates among its investors, has previously tested the system at the US Department of Homeland Security’s Transportation Security Laboratory in New Jersey, and the FBI’s headquarters in Virginia. These tests will be the first major trial of the system in real-world public environments, however.
As well as dealing with the realities of chaotic transport hubs, and tweaking the sensitivity of the devices, Evolv will need to deal with concerns about the ability of hackers to alter the algorithmic processes that underpin the detection. In the past few months, the increasing use of smart systems in transport has proven to be a cybersecurity risk, with everything from cars to rail networks being upended. You have to ask yourself: is it a good idea to place too much trust in AI security systems, if those systems are designed to be indiscernible to human security?