The AI Act: Securing Borders, Not Rights

How the AI Act Fails to Protect People on the Move and Strengthens Fortress Europe

The EU AI Act fails to protect people on the move by exempting border zones and detention facilities from key safeguards, enabling unchecked AI-driven surveillance that criminalises migrants and asylum seekers without due process. This reflects a broader trend of prioritising security and industry interests over human rights, with Greece playing a central role in weakening protection standards and expanding AI-based border technology.

For years, the EU has enhanced its border security through advanced surveillance technology, including monitoring systems, satellites, and drones, creating so-called ‘smart borders.’ More recently, AI has become integral to this security framework. As of 2 February 2025, EU member states must apply Chapters I & II of the AI Act, which regulates AI use in ‘publicly accessible spaces’ to limit civilian surveillance. However, the Act explicitly states that “publicly accessible spaces should not include prisons or border control” (EU AI Act, Recital 19). This exemption allows the unregulated use of AI in border zones and thus disproportionately targets people on the move. AI tools like Remote Biometric Identification (RBI) systems collect and store migrants’ data without evidence, suspicion, or legal justification.

A 2024 report by PICUM details the numerous risks the AI Act poses to people on the move. A key concern is the EU’s refusal to prohibit AI-driven practices that violate EU law and human rights. The Act does not ban AI-based analyses or tools that facilitate illegal practices such as pushbacks, despite evidence of their harmful impact. Frontex’s Fundamental Rights Officer warned that AI-powered surveillance could further restrict access to asylum by creating additional barriers at EU borders.

AI and Border Security: An Expanding Industry

The EU’s research agenda and budget reflect its commitment to strengthening border surveillance through AI. The EU-commissioned 2020 study Opportunities and Challenges for the Use of Artificial Intelligence in Border Control, Migration and Security highlights the EU’s aim to optimise border control using AI. Additionally, as revealed by Statewatch, the EU’s 2021–2027 financial framework increased funding for border policies by 94% (€113.3 billion) compared to the previous budget cycle. While only a fraction of this budget is allocated to border technology, let alone AI, billions of euros are available to states and companies developing such technologies.

Greece: Advocate and Laboratory for AI-Driven Border Control

Greece played a major role in shaping the AI Act’s exemptions. During EU negotiations on 29 November 2023, the Greek delegate demanded that real-time AI tools such as RBI should be used without restrictions in border zones and prisons. Greek authorities have not clarified whether Closed Control Access Centres (CCACs), Reception and Identification Centres (RCIs), or detention facilities fall under the exemption, but evidence suggests an intention to expand AI surveillance to these spaces. Shortly before pushing for the exemption in November 2023, Greek authorities had already deployed AI technologies at the Samos CCAC, as revealed by a January 2025 Solomon investigation. Correspondingly, Statewatch’s Follow the Money” investigation identifies Greece as a major testing ground for these technologies, receiving over €1 billion in EU funding for border control between 2021 and 2027 - a 248% increase from the previous budget cycle.

AI Surveillance Projects: ROBORDER and REACTION

The first AI-oriented border security project deployed in Greece was ROBORDER, which developed “unmanned mobile robots, including aerial, water surface, underwater, and ground vehicles that will operate both independently and in swarms.” Meetings between the ROBORDER research team and Greek navy officers suggest plans to implement these systems along Greek borders. The Greek Centre for Research and Technology Hellas (CERTH) was tasked with the coordination of ROBORDER. However, inquiries made to CERTH by Algorithm Watch asking for the results of the project or the utilised components were left unanswered. This lack of transparency suggests that CERTH’s work on border security would spark fierce criticism or cause exposure if made public.

Since 2021, the successor project REACTION has been reinforcing EU border security by developing AI-driven drones and automated monitoring systems. Thus, it enhances the capacity of border and coast guards to detect and prevent border crossings as the system’s algorithm learns where to focus and how to operate, allowing authorities to cover vast areas without human control. CERTH, which is also coordinating REACTION, is developing advanced computer vision and algorithms that contribute to obtaining real-time information on border crossings. REACTION is integrated into existing infrastructures such as EUROSUR and linked to information platforms at RCIs and pre-removal detention centres to facilitate control measures. In sum, with adaptive algorithms and AI-based equipment, REACTION would significantly increase the capacity of authorities to migration control, particularly in remote and vast border regions. Its full deployment is anticipated in 2025.

EU-Wide Expansion of AI Border Surveillance

Greece is not alone in pushing for expanded surveillance. According to Investigative Europe, France and other member states intended to weaken the AI Act, advocating for real-time public surveillance, AI-based border monitoring, and facial recognition profiling linked to political or religious characteristics. These amendments took effect on 2 February 2025. Human rights groups argue that the revised legislation prioritises security and industry interests over civil liberties, enabling AI-driven border enforcement with little oversight.

Frontex has also actively sought AI-based border control technologies. On 20 February 2025, it hosted an Industry Day, inviting companies to showcase AI-driven solutions for border security. This event underscores the EU’s commitment to integrating AI into its migration management and pursuing security interests at the expense of human rights.

The Need for Stronger Protection Standards

The EU AI Act contains critical gaps in safeguarding the rights of people on the move, particularly due to Greece’s efforts to exempt border zones and detention centres from key protections. The unchecked use of AI-driven surveillance risks criminalising people on the move and restricting their right to seek asylum. As several EU states continue expanding surveillance powers at the expense of human rights, urgent action is needed to advocate for stronger legal protections, transparency, and oversight to prevent the misuse of AI against people on the move.

Words by Leonard Kersting