The EU is currently running tests on iBorderCtrl, its AI lie detection program. The program will run for six months at four border points in Greece, Hungary and Latvia with other countries outside the EU.
Essentially, iBorderCtrl is a program that uses artificial intelligence tech to expedite screening processes and border crossings for travelers.
The project coordinator George Boultadakis of European Dynamics in Luxembourg tells the European Commission, “We’re employing existing and proven technologies—as well as novel ones—to empower border agents to increase the accuracy and efficiency of border checks”
“iBorderCtrl’s system will collect data that will move beyond biometrics and on to biomarkers of deceit.”
The AI system will require travelers to fill out an online application and upload identification documents (passports, ID, etc.). After that, a virtual border guard will ask questions.
According to New Scientist, some of the questions were: “What’s in your suitcase? If you open the suitcase and show me what is inside, will it confirm that your answers were true?”
Travelers will provide answers to a webcam and the AI system will determine whether they are lying by analyzing 38 micro gestures, and recording their response. This “virtual agent” will take into consideration the person’s respective gender, ethnicity, and language.
If iBorderCtl deems the passenger’s response to be truthful, they will receive a QR code which allows them to cross the border. If the traveler’s responses arouse suspicion, they will be required to provide biometric information. This procedure is rather extensive and includes finger-printing, palm vein reading, and face matching. Human personnel would later examine the data.
At the moment, the program remains strictly experimental and experts claim that it doesn’t yet have the potential to prevent someone from crossing the border. When its predecessor was tested, it only had a 76% success rate. But Keeley Crockett, a member of the iBorderCtrl team, told New Scientist that they are “quite confident” that they could raise these figures up to 85%.
There remain some concerns that if this program isn’t 100% reliable, law-abiding travelers may be misidentified as “liars”. According to the European commission, over 700 million people move in and out of the EU, a number far too great to be accommodated by iBorderCtrl AI software. Also, many studies claim that facial recognition algorithms are extremely flawed, plagued by technical issues and biases. Lastly, civil rights groups such as ACLU’s Border Litigation Project worry that the AI software could infringe on individual privacy.
Among these concerns, it is slightly reassuring to know that program is only being implemented in select countries for a limited trial period. It is imperative that the system obtains as much training data from as diverse a pool of travelers as possible.
In a time when crossing borders is already a contentious issue, it seems that adding a flawed “impartial agent” is understandably a concern.