A GenAI-Powered Cybersecurity Mesh for Real-Time Risk Detection in Digital Payments

The surge in digital payment systems has introduced significant challenges in securing real-time financial transactions against sophisticated cyber threats. This paper presents a GenAI-powered cybersecurity mesh architecture designed to detect and mitigate risks dynamically within decentralized digital payment ecosystems. Our approach integrates transformer-based deep learning models, explainable AI (XAI) mechanisms, and decentralized threat telemetry across a microservices-driven architecture. The system supports autonomous policy enforcement and real-time fraud detection using SHAP-explainable risk scores aggregated from user behavior, device fingerprinting, and transaction metadata. Deployed within a zero-trust security model, the mesh adapts to adversarial patterns through self-improving AI agents. Experimental results on simulated fintech datasets show over 96% detection accuracy with sub-200ms latency, demonstrating viability for high-throughput environments. The proposed architecture advances the state-of-the-art in real-time payment security and paves the way for scalable GenAI-driven threat response infrastructures in financial services.