AI Risk Management for Financial Institutions: Building Defensible Governance for Vendor-Driven AI (en Inglés) - Barry A. Salas

AI Risk Management for Financial Institutions addresses a growing yet often overlooked challenge: artificial intelligence is already deeply embedded across financial institutions, frequently without formal governance, oversight, or structured risk controls. As vendor-driven AI solutions continue to expand, credit unions and similar institutions are increasingly exposed to regulatory, operational, and reputational risks—often without full visibility or awareness.

This book offers a practical, implementation-focused framework for understanding and managing AI risk in real-world financial environments. Beginning with the “quiet arrival” of AI, it examines how vendor-embedded models, shadow AI usage, and passive adoption contribute to critical governance gaps. Readers are introduced to a structured approach to AI risk, encompassing opacity, data dependency, model drift, and institutional reliance, along with the AI Governance Pyramid Framework to help assign responsibilities across teams. The book also aligns AI risk management with existing regulatory expectations, including SR 11‑7, FFIEC guidance, data protection requirements, and fair lending obligations. In addition, it presents practical methodologies for AI risk classification and pre-deployment evaluation, enabling institutions to assess systems before they are placed into production.

By bridging the gap between theory and execution, this book equips leaders, risk professionals, and boards with actionable tools to progress from awareness to defensible AI governance—supporting responsible, transparent, and regulator-aligned AI adoption.

What you will learn:

Who this book is for:

This book is intended for risk, compliance, audit, and governance professionals within regulated financial institutions that rely on vendor-driven AI systems. It is also suited for technology leaders, data governance teams, and AI oversight committees responsible for evaluating and managing AI risk.