Document Type
Book Chapter
Publication Title
Insights in Banking Analytics and Regulatory Compliance Using AI
Abstract
This chapter develops an analytical model for Explainable AI (XAI) designed to support major types of banking regulatory compliance audits. Financial institutions increasingly rely on AI for efficiency in compliance tasks, but the opacity of complex models creates challenges in meeting regulatory transparency requirements. The proposed mathematical framework incorporates explainability as a quantifiable constraint in the optimization of AI models across five critical audit categories: CAMELS examinations, BSA/AML compliance, Consumer Compliance, IT Audits, and Internal Audits. By quantifying explainability through information- theoretic measures and Shapley values, the model balances predictive performance with regulatory requirements. The methodology is validated through a numerical example demonstrating significant improvement in the objective function while satisfying all regulatory constraints.The chapter addresses implementation challenges across different regulatory jurisdictions, human- AI collaboration considerations, and resource requirements for institutions of varying sizes.
First Page
259
Last Page
284
DOI
10.4018/979-8-3373-0209-6.ch013
Publication Date
2025
ISBN
9798337302096
Recommended Citation
Desai, Hrishikesh, "Reimagining Compliance: Explainable AI Models for Financial Regulatory Audits" (2025). Faculty Publications. 32.
https://arch.astate.edu/busn-isba-facpub/32
