This article introduces a novel architecture that combines temporal graph neural networks (TGNNs) with explainable AI techniques to detect, visualize, and remediate compliance policy drift as it happens. Readers will learn the data pipeline, model design, interpretability methods, and how to embed the solution into CI/CD and governance workflows, all illustrated with Mermaid diagrams and practical implementation tips.