AI Resilience Report forFinancial Examiners

Right now, AI is showing up on both sides of the examiner's desk — as a tool examiners use and as a subject they investigate. In central banks around the world, supervisors are using AI to gain "more reach, speed and precision" as the financial system grows more complex, with AI applications enabling a shift to a proactive supervisory approach where AI flags anomalies for human experts to investigate, significantly streamlining regulatory compliance monitoring and enhancing risk assessment capabilities. On the industry side, banks are automating Sarbanes–Oxley controls, BSA/AML monitoring, and risk detection ^[1], which changes what examiners find when they review balance sheets and loan files.

At the same time, examiners have a brand-new job: policing AI itself. As of mid-2026, U.S. banking regulators have begun incorporating artificial intelligence oversight into every routine bank examination, pressing lenders on how they deploy the technology across higher-risk functions including lending, customer verification, and sanctions screening, and the Federal Reserve's revised model risk guidance in SR 26-2 ^[2] gives examiners a new rulebook for judging bank AI systems. So the routine "read the documents" tasks are being augmented, while the judgment-heavy investigative tasks are actually expanding.

Adoption in supervision is moving carefully rather than quickly. CSBS Chair Rhoshunda Kelly told state and federal supervisors that guidance will be critical as banks look to incorporate new tools, like artificial intelligence, into their business model, and that the consequences if banks are not allowed to innovate could be dire — signaling regulators want AI in the system but on their terms. Two forces speed adoption: the labor market is tight, since employment of financial examiners is projected to grow 9 percent from 2025 to 2035, much faster than the average for all occupations, with about 4,500 openings projected each year, and AI vendors promise that regulatory capacity can increase without a proportional rise in compliance cost or headcount ^[1].

Two forces slow it down: legal accountability (an examiner's signature carries weight a chatbot's cannot) and the need for auditable, explainable decisions. The hopeful takeaway for students: the tasks with the highest automation scores — reading documents and drafting reports — are the ones AI helps with, while the human parts (meeting with directors, exercising legal judgment, supervising junior staff) remain firmly yours to grow into.