Application of artificial intelligence and machine learning algorithms as a tool for combating tax and customs fraud in the EU and Serbia

Abstract

The development of artificial intelligence (AI) and machine learning (ML) technologies has significantly influenced the modernization of tax and customs administrations. Modern administrations increasingly rely on digital tools for processing large volumes of data, risk assessment, fraud detection, and communication with taxpayers and customs users. The application of AI and ML systems enables faster identification of suspicious transactions, automated verification procedures, and more efficient tax and customs control. This paper examines the application of artificial intelligence and machine learning algorithms in tax and customs administrations, with a particular focus on selected European Union member states and Serbia. Special attention is given to the use of risk-scoring systems, automated document processing, AI chatbots, cargo monitoring systems, and digital customs solutions. The paper also highlights the advantages and limitations of AI implementation in procedures related to the prevention of tax and customs fraud. The analysis shows that EU member states have already introduced various AI-based solutions into tax and customs operations, while Serbia has gradually started implementing similar technologies through projects related to risk analysis, e-fiscalization, digital customs systems, and virtual taxpayer assistance. Despite the significant potential of artificial intelligence, its effective application still depends on data quality, transparency, legal certainty, and human supervision.