<p>This study examines the governance of artificial intelligence (AI) in the financial sector with a specific focus on Saudi Arabia. Drawing on a three-theory analytical framework—Institutional Theory, Agency Theory, and Risk Governance Theory—the paper develops a structured conceptual model that explains how AI adoption interacts with regulatory structures, accountability relationships, and risk management systems. Each theoretical lens is applied to a distinct governance dimension: Institutional Theory explains why regulatory frameworks lag behind AI adoption; Agency Theory explains how distributed decision-making in AI systems creates accountability gaps; and Risk Governance Theory explains why existing risk management frameworks are inadequate for dynamic, self-learning systems. The paper conducts a comparative analysis of AI regulatory approaches in the European Union, the United Kingdom, and the United States, drawing substantive governance lessons for the Saudi context. It identifies five specific governance gaps in the Saudi financial regulatory landscape and develops ten Saudi-specific policy recommendations grounded in the Saudi institutional architecture—including SAMA, SDAIA, CMA, and the Vision 2030 regulatory environment. An illustrative case using actual classification outputs from a machine learning model demonstrates the governance interpretation challenges that technical accuracy alone cannot resolve. The study contributes a theory-differentiated analytical framework for AI governance in financial systems, advancing beyond descriptive policy analysis to provide structured, actionable governance guidance for an emerging economy undergoing rapid digital transformation.</p>