US Bank Examiners Receive New Guidance

American banking supervisors have received their first dedicated framework for assessing artificial intelligence, with regulators emphasising the risks created by how banks deploy technology rather than treating AI as an entirely separate category of financial risk.

The Office of the Comptroller of the Currency issued the guidance on Wednesday, directing examiners to consider the scale, complexity and consequences of AI applications used by national banks. The framework covers advanced systems capable of performing tasks with limited human intervention, including AI agents that can initiate financial transactions or support operational decisions.

Regulators have been examining how banks use increasingly sophisticated models for customer service, fraud detection, lending and internal operations. Under the new approach, a relatively simple automated system authorised to move money could warrant closer scrutiny than a more advanced model used for low-risk administrative work. Examiners are encouraged to consider human oversight, restrictions on autonomy and the potential consequences of errors. The OCC stressed that its framework does not impose new legally enforceable requirements or oblige banks to establish separate risk limits specifically for AI.

The guidance signals an effort to accommodate technological innovation without abandoning established principles of banking supervision. For financial institutions, it reinforces the importance of understanding what automated systems can do and who remains accountable when they fail. As AI moves from assisting employees towards executing transactions and other consequential tasks, the distinction between technical sophistication and operational risk may become increasingly important to regulators.