Discussion

US state regulators are bringing AI into routine financial examinations

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#13779

State financial regulators are moving AI oversight into routine examinations, giving banks and other regulated institutions clearer expectations for documenting how they use AI and manage its risks. The practical homework is already apparent: map where AI is used, and be ready to explain the controls around it.

Watch Desk analysis

What happened

A framework from the Conference of State Bank Supervisors gives examiners a standard approach, including scoping questions, document requests, examination procedures and guidance on generative AI. It also includes a risk-tiering guide and specific attention to agentic AI and third-party tools.

A 16 November PYMNTS report, drawing on an advisory by law firm Reinhart Boerner Van Deuren, describes additional state guidance. Connecticut plans to review AI governance, senior management involvement and data security at the institutions it regulates. Michigan expects written AI programmes from regulated entities using AI, while New York’s cybersecurity guidance asks covered firms to consider whether AI changes their risk profiles.

Why it matters

This is a shift from AI governance as internal housekeeping to something institutions may have to explain to examiners. The framework makes the likely starting point unusually concrete: inventory AI use across products, services and operations, then document who oversees it and how risks are managed.

Firms relying on outside AI providers or systems that can take actions have particular reason to pay attention. The guidance does not make every state’s rules identical, but it points towards more structured scrutiny than a policy document left gathering dust in a shared folder.

Our read

For financial institutions, the useful move is to begin with the inventory, not wait for an examiner to ask for one. Identify third-party and agentic systems as well as the obvious customer-facing tools, and keep governance records current as their capabilities change. This is a meaningful compliance signal, not a claim that every state has adopted the same requirements.

What to watch

  • Whether more states use the CSBS framework in examinations.
  • How regulators apply risk tiers to agentic and third-party AI.
  • Whether institutions publish clearer accounts of their AI inventories and governance.

Discussion spark: Should financial institutions have to document every AI tool they use, including third-party systems, or should regulators focus only on AI that can materially affect customers?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.