Industrial Automatic Control Systems and Controllers Annotation << Back Graph Neural Network Algorithms in Anti-Money Laundering (AML): Emerging Challenges and Regulatory Aspects Putkov K.A., Ovsyannikov R.Yu. The introduction of artificial intelligence technologies, particularly graph neural networks (GNNs), opens new possibilities for detecting complex multi-level money laundering schemes that remain invisible to traditional rule-based systems. However, the application of deep learning in anti-money laundering (AML) is fraught with fundamental challenges: the problem of model interpretability ("black box") and the need to comply with stringent regulatory requirements for transparency and accountability. This article examines the potential of GNNs for analyzing fi nancial transaction networks, as well as explainable artificial intelligence (XAI) methods as a key solution to ensure regulator trust. The study demonstrates that a hybrid GNNXAI approach can radically reduce the false positive rate by 60–80 % while simultaneously improving detection accuracy and providing the transparency necessary for audit. Practical recommendations for integrating these technologies in line with the risk-based approach enshrined in FATF and Bank of Russia documents are formulated. Keywords: anti-money laundering (AML), graph neural networks, explainable artificial intelligence (XAI), graph analytics, regulatory requirements, false positives, suspicious transactions. DOI: 10.25791/asu.3.2026.1648 Pp. 36-42.