EU projects explain how to create ethical AI at the border

The European Union’s AI Act will have wide-reaching effects across various industries. One of the areas where AI use will be regulated is something most of us have experienced  – border crossings.

The EU is currently testing new border technology through several innovation projects, including those that rely on biometrics. Among them are Horizon Europe-funded Odysseus and FlexiCross which explained during a recent workshop how they are integrating the AI Act regulation and managing ethical and societal concerns in border management security technologies.

“It’s still quite early to say how the implementation will go,” says Dana Oniga, project manager at Romanian software company Simavi.

Both projects are currently testing border technologies at the borders of countries such as Greece, Romania, Moldova and Bulgaria.

Odysseus is designed to make border crossings easier through seamless identity verification, digital and virtual passports, face matching, behavioral authentication and other technologies such as drones. The project involves 14 partners from 12 European countries, including companies such as Simavi and biometric technology providers Thales and Vision-Box.

Similarly, FlexiCross (Flexible and Improved Border-Crossing Experience for Passengers and Authorities) aims to boost border security while streamlining management with tools such as predictive risk assessment, biometric checks, data exchange, human-machine interfaces, augmented reality for situational awareness and more.

The project’s task is to deliver a toolkit comprising at least 20 modules. Its ethics risk analysis, however, showed plenty of areas that need to be improved, according to Ioana Cristina Cotoi, legal researcher at Engineering Ingegneria Informatica SPA.

The evaluated risks included everything from bias and unjust treatment in biometric recognition and unusual pattern detection systems to electrocution from malfunctioning drones or biometric systems.

“Another risk that we have identified is the increased stress and anxiety among travelers due to constant surveillance and the use of advanced monitoring systems such as drones, fingerprint scans and other biometric checks at border crossings,” says Cotoi.

Data security and safety emerged as the predominant risk, followed by concerns over surveillance, risk to health and safety and eventually violation of traveler rights such as freedom and equality. FlexiPass has come up with a range of strategies to mitigate them.

According to their analysis, bias should be addressed by ensuring diversity in data collection during the algorithm training, refining detection algorithms and providing alternative screening options. Surveillance or inappropriate use of technology, particularly biometric recognition should be mitigated by enforcing strict access control.

Finally, travelers should be offered support when stressed while border authorities should implement a clear and transparent border check process. Surprisingly, another mitigation measure was reducing the use of technologies whenever possible.

Odysseus, on the other hand, has introduced a procedure for legal analysis and risk-level categorization using EU tools such as the EU AI Act Compliance Checker and ALTAI (Assessment List for Trustworthy Artificial Intelligence). Following the assessments, the project will introduce compliance measures.

“There are a lot of questions that need to be posed when we talk about ethical aspects, says Oniga. “There are rule of law, application, explanation, testing and validations that needs to be done, quality of the service.”

Article Topics

biometric identification  |  biometric-bias  |  biometrics  |  border security  |  data protection  |  ethics  |  EU  |  explainability  |  FlexiCross  |  Odysseus