The Deposit Report
Analysis

OCC tells examiners to probe your AI inside existing risk categories, not a new one

The OCC issued an examiner framework for bank use of what it calls "super intelligence." You will not need a separate AI risk appetite, but examiners can tailor their work to how you use the technology in credit, payments, fraud detection and customer service.

You do not have to write a standalone AI risk appetite statement for your board. The OCC said so this week, and in the same breath told its examiners to go find your AI inside the risk programs you already run. (OCC, November 4, 2026; OCC)

The ABA Banking Journal reported the release the same day. The guidance went out as OCC Bulletin 2026-54, addressed to the chief executives of national banks, federal savings associations and federal branches and agencies. (ABA Banking Journal, November 4, 2026)

The OCC says AI is not a separate risk category

The single line to read twice is this: examiners "should not treat SI as a separate and distinct risk category, and banks should not be required to define a separate SI risk appetite," the agency said in its November 4 release. That means no standalone AI risk appetite statement for your board to write and your examiner to grade. Instead, the bulletin says the framework "may be used by examiners to identify SI-specific considerations that supplement existing model risk management, governance, technology, compliance, and operational-risk practices". (OCC, November 4, 2026; ABA Banking Journal, November 4, 2026; OCC)

So the AI questions land inside the programs you already have on the books. The ABA Banking Journal reported that examiners will use the framework when evaluating AI use in "activities such as credit decisions, payments, fraud detection, customer service, operations and risk management". If you use AI in any of those, that is where an examiner will look, through the risk-management framework that already covers it.

The framework sets no new rule, but it shapes the exam

The bulletin is blunt about its own limits: it "does not establish supervisory expectations for banks" and is "intended as a resource to assist examiners in their supervisory activities," the OCC said on November 4, 2026. It also says nothing in the guide "is intended to establish new regulatory expectations or set forth enforceable standards or prescriptive requirements". (OCC)

It is still the script your examiner will carry. The OCC instructed examiners to "tailor their examination of banks' use of SI based on the materiality of the SI use case," in a bulletin signed by James M. Gallagher, Senior Deputy Comptroller and Chief National Bank Examiner (OCC). The bulletin also says a bank's risk classification may rest on "potential consequences, transfer of agency, and the degree of human oversight rather than the technical sophistication of the SI used". In plain terms, how much a tool can do on its own, and how little a human checks it, matters more than how advanced it is.

What this means for a community bank

The OCC wrote a note for community banks into the bulletin. It instructs examiners to "tailor their examinations to the bank's size, complexity, and risk profile," proportionate to "the nature, consequences, autonomy, and complexity of the bank's SI use cases". The bulletin also says a bank's "control intensity could increase with the potential consequences of an SI agent's failure or misuse". (OCC)

Read together, the examiner's first question will not be whether your AI is sophisticated. It will be what each use case could do on its own, what happens if it fails, and who checks it, because those are the things the bulletin says drive the risk classification. The OCC does not name the tools, so you have to map your own before the next exam. The framework will be regularly updated, the OCC said, so whatever you write down now you will have to keep current. (ABA Banking Journal, November 4, 2026; OCC, November 4, 2026)

What to do this month

  1. Decide, before your next exam date is set, which of your AI use cases are material, because the OCC told examiners to tailor the exam to the materiality of each use case. Make that call now, not in the exam room.
  2. List every AI or agentic tool in credit decisions, payments, fraud detection and customer service, since the ABA Banking Journal reported those are the activities examiners will evaluate. For each one, write down the potential consequences, the degree of human oversight and whether the tool acts on its own, the three things the OCC named as the basis for risk classification.
  3. Ask any vendor whose product uses AI, in writing, to describe what the tool decides without a human, so you can match it to the OCC's test of transfer of agency and human oversight before an examiner asks you.

Methodology

This article is built from published reporting and primary documents, not from Ledger data. The primary documents are OCC news release 2026-97 and OCC Bulletin 2026-54, both dated November 4, 2026; the trade reporting is the ABA Banking Journal's writeup from the same day. The documents set no new rule and describe no specific bank incidents, and they do not give a count of how many banks use AI today.

  1. OCC, 'OCC Issues Supervisory Framework on Bank Use of Super Intelligence', November 4, 2026.
  2. OCC, 'Super Intelligence: Supervisory Framework'.
  3. ABA Banking Journal, 'OCC releases framework for supervising artificial intelligence', November 4, 2026.