Proceedings of the International scientific and practical conference ―Science, Technology and Culture: Dynamics of Change in the XXI Century‖ (December 1921, 2025) / Publisher website: www.naukainfo.com. – Baku, Azerbaijan, 2026. – 90 p.

21 INFORMATION TECHNOLOGIES AND SYSTEMS UDC 004.8 Honcharuk Oleh Mykolayovych Candidate of Philological Sciences, Associate Professor, Professor of the Department of Language Training Kharkiv National University of Internal Affairs Voropaiev Denys Vitaliiovych Didenko Oleksandr Romanovych Sitalo MaksymAndriyovich Cadets of Educational and Scientific Institute No. 4 Kharkiv National University of Internal Affairs, Kamianets-Podilskyi, Ukraine DEVELOPMENT OFAMECHANISM FOR PROTECTION AGAINST DATA LEAKAGE ATTACKS USING FEDERATED LEARNING AND DIFFERENTIAL PRIVACY Federated Learning (FL), in combination with Differential Privacy (DP), has become a key research direction amid the rapid development of digital technologies and machine learning. As the scale of data processing increases, so does the need to ensure information confidentiality, especially when dealing with sensitive data such as medical records, financial documentation, or users‘ personal information. Traditional centralized approaches to training machine learning models are associated with significant risks of data leakage, which exacerbates security concerns. Even intermediate information transmitted during training (for example, model updates)