Biography
Andrej Mitić, PhD, is a Lecturer in Law and Artificial Intelligence at the Academy of Applied Technical and Preschool Studies Niš, Pirot Department, Serbia. He holds a PhD in Law from the University of Niš and a degree in Philosophy from the University of Belgrade, and was a postdoctoral fellow at New Europe College in Bucharest and at Sofia University. He holds certificates in AI ethics, regulation, and compliance from Saïd Business School, University of Oxford, and is a Certified AI Ethicist (CAIE, Oxethica). His research focuses on AI Ethics, regulation, and compliance, the EU AI Act, and AI governance in candidate countries. He is a member of the Italian Society for the Ethics of AI (SIpEIA).
Conference contributions 1
THE PRE-ACCESSION “BRUSSELS EFFECT”
Transposing the EU AI Act in the Western Balkans
The EU AI Act is expected to reproduce the "Brussels effect" (Bradford) that turned the GDPR into a global standard. The literature describes this effect as market-driven. This paper argues that in EU candidate countries the mechanism is different: the AI Act is adopted by accession obligation, often before the institutions needed to apply it exist. This could be called a pre-accession Brussels effect, in which the legal text travels faster than regulatory capacity. The GDPR offers a precedent. Data protection laws in Serbia and North Macedonia, closely modelled on it, show the limits of near-literal transposition and the gap between law in the books and law in action. The AI Act raises the stakes, as it depends on harmonised standards, notified bodies and market surveillance authorities that candidate countries largely lack. It is also a moving target: with the Digital Omnibus (2026), the EU itself postponed most high-risk obligations to 2027 and 2028. The paper compares Slovenia and Croatia, which apply the Act directly as Member States, with Serbia, which is preparing the region's first comprehensive AI law, and with Montenegro, North Macedonia and Albania, which have no AI-specific legislation yet. The paper argues for a flexible model of reception that distinguishes text transfer from capacity transfer: alignment of the core rules and implementation synchronised with the EU timeline.