The IEAI discusses AI and border security
On January 30th, 2025, the IEAI hosted an online discussion on “AI in Border Security: Opportunities and Pitfalls” with Maria Chiara Aquilino – Junior Analyst at RAND Europe and Prof. Henri van Soest – Senior Analyst at RAND Europe and professor of Policy Analysis at RAND School of Public Policy.
Ms. Aquilino began by first providing a background on RAND’s work and roles of the key actors in EU border security, including the European Border and Coast Guard Agency (Frontex), national border agencies, other EU agencies and international agencies. Frontex´s main aims include border surveillance, fighting cross-border crime and monitoring migratory flows. It also works to modernize border control processes and implement new technologies to improve efficiency and security. Due to an upward trend in recent years in border crossings from the East and from Africa, the EU has equipped its border infrastructure with advanced surveillance technologies and autonomous systems while collaborating with non-European countries to manage migration flows.
Prof. van Soest then explained how AI has been applied in border security in more detail. The earliest foundations were laid in 2013 with the introduction of Smart Border Package. Since 2015, AI in border security has gone through two phases of technology development. First from 2015 – 2020, research was mainly concentrated on large scale information systems, biometric data collection and border control technologies. In a second phase, known as advanced implementation (2020 – 2025), focus was shifted to the widespread deployment of AI technologies. Some prominent examples are automated fingerprint or face recognition, algorithmic risk assessment and migration monitoring and forecasting. At the moment, ongoing research is being conducted on emotion detection and autonomous border security applications.
Prof. van Soest categorised the use of AI in border security into five main areas: 1) situation awareness and assessment, 2) information management, 3) communication, 4) detection, identification and authentication, and 5) training and exercises.
As AI is still at its early stages, pitfalls are inevitable according to our speakers. Technically, risks associated with data privacy, cybersecurity, black box problems and false positive or negative identifications have yet to be completely addressed. Operationally, end users tend to have an inadequate understanding of AI, leading to overreliance on technological solutions. Therefore, knowledge gaps in understanding these tools should be a goal for European border agencies.
In terms of the legal and regulatory setting, current governance frameworks have yet to be specific enough in the border security context and should aim to balance security needs with individual rights. Prof. van Soest emphasized ethical concerns citing human rights violations, discriminative decision-making, dehumanized migration processes and biases. Because of this, comprehensive risk assessment with ethical dimensions and human rights should be the precursor for AI applications.
We thank Ms. Aquilino and Prof. van Soest for their intriguing and informative presentation on AI in border security. The recording of the event can be found here.
