From Cryptocurrencies to Collaborative Risk Management: A Review of Decentralized AI Approaches

Abstract

Enterprises today face increasing threats from cyberattacks, supply chain disruptions, and systemic market risks, making the enhancement of organizational resilience through advanced risk management frameworks increasingly critical. Traditional approaches often struggle to balance data privacy, cross-organizational collaboration, and real-time adaptability. While distributed ledger technologies (DLTs) initially enabled cryptocurrencies, they have evolved into a foundational infrastructure for decentralized AI applications. This study investigates how decentralized AI techniques, particularly federated learning, can support joint risk management processes in enterprise networks. First, a comprehensive review of decentralized AI methods is conducted to identify approaches suitable for enterprise risk management. Next, expert interviews are used to contextualize these insights, highlighting practical considerations, organizational challenges, and adoption constraints. Building on the literature and expert feedback, a decentralized framework is developed to allow organizations to securely share risk-related insights while preserving data privacy and control over proprietary information. The framework is validated through a technical prototype, combining architectural design with empirical proof-of-concept experiments on federated learning benchmarks. Results demonstrate the feasibility of achieving near-centralized model accuracy under privacy constraints, while also highlighting communication and governance issues that need to be addressed in real-world deployments. The study presents a structured comparison of decentralized AI techniques and a validated concept for enhancing supply chain risk prediction, fraud detection, and operational continuity across enterprise networks.