Authors: Meena P. Subramanian
Abstract: As global enterprises increasingly rely on artificial intelligence (AI) to drive decision-making, they face growing challenges related to data sovereignty, privacy, and regulatory compliance. Traditional AI models rely on centralized data aggregation, often violating regional data protection laws such as GDPR, PDPB, and China’s Data Security Law. Federated AI—a decentralized learning approach—has emerged as a solution that enables organizations to train AI models collaboratively without transferring raw data across borders. This review explores how federated AI influences data sovereignty in global enterprises by balancing innovation with compliance. It presents the underlying principles of federated learning, detailing its architecture, operational workflow, and privacy-preserving mechanisms. The analysis highlights how federated AI ensures compliance through decentralized data governance, secure aggregation, and encryption-based privacy protection. It further discusses regulatory alignment across jurisdictions and real-world applications in sectors such as healthcare, finance, and telecommunications. The paper also identifies major challenges including communication overhead, data heterogeneity, model inversion risks, and the absence of global interoperability standards. Comparative analysis demonstrates that while centralized AI offers efficiency and simplicity, federated AI provides superior compliance, resilience, and user trust—key attributes for multinational enterprises operating under diverse legal frameworks.
DOI: http://doi.org/10.5281/zenodo.17799859
