Federated Learning and Differential Privacy in Multi-Cloud Banking
Main Article Content
Abstract
By 2024, the financial services sector has reached an inflection point where artificial intelligence (AI) has become indispensable for fraud detection, credit risk modeling, anti–money laundering (AML), and real-time analytics. Yet, the same period has also seen tightening global expectations for privacy protection, data residency, and multi-jurisdictional compliance. As banks move toward multi-cloud strategiesdistributing regulated workloads across platforms such as Amazon Web Services (AWS), Microsoft Azure, and private sovereign clouds-the inherent fragmentation of data creates tension between analytics ambitions and legal constraints. Federated Learning (FL) and Differential Privacy (DP) offer a viable technical reconciliation, enabling cross-institutional intelligence without compromising privacy or regulatory boundaries. This paper analyzes the conceptual foundations, system architectures, privacy guarantees, and strategic implications of FL and DP within multi-cloud banking ecosystems. It synthesizes relevant 2018–2024 research and provides a practice-oriented interpretation grounded in the banking sector’s current technological and regulatory reality.