This article critically examines the deployment of Artificial Intelligence (AI)-based pre-screening systems in the European Union’s (EU) border and migration management, with a particular focus on the human rights implications under the 1951 Refugee Convention, the EU Charter of Fundamental Rights, and the European Convention on Human Rights (ECHR). The study analyses how automated risk assessment models, biometric data mining, and linguistic analysis tools—such as the “Centaur” project in Greece and the “LINGUA” pilot in Belgium—affect asylum seekers’ procedural safeguards. A central concern identified is the prevalence of algorithmic bias and discrimination (Korbayram, A., & Hoca, E., 2024a). Historical migration data embedded with geopolitical and ethnic prejudices often result in disproportionate risk classifications, particularly disadvantaging applicants from specific countries. This entrenches unequal treatment in violation of Article 21 of the EU Charter (non-discrimination) and undermines Article 41 (right to good administration). Furthermore, the opacity of “black box” algorithms inhibits applicants’ ability to contest decisions effectively, raising concerns under Article 13 ECHR and the obligation for individualised assessment in refugee law. Through case studies of Greece and Hungary, the article illustrates how flawed AI outputs have led to wrongful returns, prolonged detention, and compromised living conditions in camps. These outcomes reveal the tension between technological efficiency and the absolute prohibition of torture or inhuman treatment under Article 3 ECHR. The paper proposes a rights-by-design approach, incorporating Algorithmic Impact Assessments, independent bias audits, and strengthened appeal mechanisms. It also calls for AI Act migrationspecific protocols, greater civil society oversight, and hybrid decision-making models where human officials retain ultimate veto power. The conclusion underscores that “smart borders” must balance operational gains with the imperative to protect human dignity, ensuring AI functions as a tool for enhancing—rather than eroding—fundamental rights.
Dioan GEGAJ (Tue,) studied this question.