How does the OCC regulate AI in banking?

Last updated Nov 17, 2026 · Updated as rules change

The OCC supervises AI at national banks through model risk management, safety-and-soundness examination and, since November 4, 2026, a dedicated examiner framework — not through AI-specific rules. On April 17, 2026 the OCC, Federal Reserve, and FDIC issued revised interagency model risk management guidance (OCC Bulletin 2026-13) that supersedes the 2011 framework, rescinds OCC 2011-12, and excludes generative and agentic AI from its scope. On November 4, 2026 the OCC issued Bulletin 2026-54, 'Super Intelligence: Supervisory Framework', a resource telling its examiners how to assess exactly those generative and agentic systems — sixteen components, six pillars and a four-tier, consequence-based classification — while stating that it establishes no supervisory expectations for banks.

Checked 2026-11-17 · Office of the Comptroller of the Currency
Full nameOffice of the Comptroller of the Currency
RolePrudential supervisor
Force on banksSupervisory guidance
Applies toNational banks, federal savings associations, and federal branches of foreign banks
Key documentOCC Bulletin 2026-54 — Super Intelligence: Supervisory Framework for examiners (Nov 2026); Bulletin 2026-13 for model risk (Apr 2026)
Latest moveNov 4, 2026: Bulletin 2026-54, the Super Intelligence supervisory framework for examiners on generative and agentic AI; Apr 2026 revised model risk guidance excluding those systems
Documents tracked9 · all documents →

For fifteen years the operative document was the 2011 Supervisory Guidance on Model Risk Management (OCC 2011-12 / Fed SR 11-7). The April 2026 revision modernizes that framework and makes a deliberate scoping choice: generative and agentic AI are called 'novel and rapidly evolving' and carved out of formal model-risk requirements, with banks told to govern them through enterprise risk management instead. The OCC simultaneously rescinded older issuances including its 1997 credit-scoring-models bulletin and 2021 BSA/AML model risk FAQ.

Beyond model risk, the OCC's May 2026 Semiannual Risk Perspective flags AI as a driver of fraud and increasingly fast, sophisticated cyberattacks. The gap the April carve-out left was filled from the examiner's side on November 4, 2026: Bulletin 2026-54, 'Super Intelligence: Supervisory Framework', adopts Executive Order 14434's term for AI and gives examiners sixteen components (purpose, risk classification, human-in-the-loop and human-on-the-loop controls, governance, model risk extensions, data, privacy, compliance, third parties, resilience, cybersecurity, monitoring, testing, documentation, audit, reporting) grouped in six pillars, plus an illustrative four-tier classification in which an autonomous payment agent is Tier 1 and an internal productivity tool Tier 4. The bulletin says the framework sets no supervisory expectations, enforceable standards or prescriptive requirements for banks and will be updated regularly. The practical posture in 2026 is innovation-friendly: examiners test whether banks understand and control their AI, not whether they use it.

What is the OCC's guidance on AI (artificial intelligence)?

OCC guidance on AI is a set of supervisory guidance, one examiner framework, a risk report and a speech; the OCC has not issued an AI-specific rule. In date order: the March 31, 2021 interagency request for information on AI (OCC Bulletin 2021-17), which led to no rule; the June 6, 2023 interagency third-party risk management guidance (Bulletin 2023-17), which banks apply to AI vendors; Acting Comptroller Hood's April 29, 2025 remarks on AI in financial services; the April 17, 2026 revised model risk management guidance (Bulletin 2026-13), which places generative and agentic AI outside its scope; the Spring 2026 Semiannual Risk Perspective (May 7, 2026); and, on November 4, 2026, Bulletin 2026-54, 'Super Intelligence: Supervisory Framework', the OCC AI framework for examiners assessing generative and agentic systems, which states that it 'does not establish supervisory expectations for banks'. The OCC artificial intelligence guidance in force for banks today is therefore Bulletin 2026-13 for models and Bulletin 2023-17 for vendors, with Bulletin 2026-54 showing what examiners look at.

RuleAuthorityWhat it requiresApplies
OCC Bulletin 2011-12 (Supervisory Guidance on Model Risk Management)OCCThe 2011 model risk management guidance, the twin of the Federal Reserve's SR 11-7, under which national banks governed their models for fifteen years. Not written for AI; replaced by Bulletin 2026-13.Apr 4, 2011; rescinded Apr 17, 2026
OCC Bulletin 2021-17 (interagency request for information on AI)OCCThe OCC, Federal Reserve, FDIC, CFPB and NCUA asked how financial institutions use AI and machine learning, and about explainability, data quality and bias, overfitting, cybersecurity, dynamic updating, third-party AI and fair lending. An information request with no obligations; no AI-specific rule followed.Mar 31, 2021; comments closed Jul 1, 2021
OCC Bulletin 2021-19 (model risk management for BSA/AML systems)OCCInteragency statement that banks could test and update BSA/AML systems, including with new technology such as machine learning, without automatic supervisory criticism.Apr 9, 2021; rescinded Apr 17, 2026
OCC Bulletin 2023-17 (interagency third-party risk management guidance)OCCThe framework banks apply to AI vendors, cloud and model providers and fintech partners: planning, due diligence and selection, contract negotiation, ongoing monitoring and termination, scaled to the risk and criticality of each relationship.In force since Jun 6, 2023; replacement proposed Sep 11, 2026
Acting Comptroller Hood, 'AI in Financial Services'OCCThe OCC expects 'robust risk management practices for AI applications, consistent with conventional model risk management practices', covering bias, security vulnerabilities and explainability. A statement of supervisory posture, not a requirement.Speech, Apr 29, 2025
OCC Bulletin 2026-13 (Model Risk Management: Revised Guidance)OCCInteragency guidance with the Federal Reserve (SR 26-2) and FDIC (FIL-15-2026): model development and use, validation and ongoing monitoring, governance and controls, and vendor models, most directly relevant above $30 billion in total assets. Generative and agentic AI are outside its scope. Supervisory guidance, not a rule.In force since Apr 17, 2026
OCC Semiannual Risk Perspective, Spring 2026OCCReports that AI is 'significantly transforming the cyber threat landscape' and that banks are taking a 'measured approach' to generative and agentic AI; says the OCC is 'actively reviewing supervisory expectations, guidance, and regulations'. Supervisory priorities, not requirements.Report, May 7, 2026
OCC Bulletin 2026-46 (proposed interagency third-party risk management guidance) ↗OCCProposal with the Federal Reserve, FDIC and NCUA to rescind and replace the existing third-party risk management guidance. The proposal does not itself address AI, but it would reset the vendor framework banks use for AI providers.Proposed Sep 11, 2026; comments due Dec 16, 2026 (extended from Nov 16 by Bulletin 2026-57)
OCC Bulletin 2026-54 (Super Intelligence: Supervisory Framework)OCCA resource for OCC examiners evaluating banks' use of 'Super Intelligence', the OCC's term for AI under Executive Order 14434, 'more applicable to generative and agentic SI': sixteen components in six pillars and an illustrative four-tier, consequence-based classification. It sets no supervisory expectations, enforceable standards or prescriptive requirements for banks.Issued Nov 4, 2026; to be updated regularly

None of these documents is a regulation. Bulletin 2026-13 and Bulletin 2023-17 are supervisory guidance applied through examinations; Bulletin 2026-54 says banks 'are not expected to use this document to assess their use of SI or their SI risk management'; the Semiannual Risk Perspective and the Hood remarks describe priorities and posture. AI at national banks remains subject to existing law, including safety and soundness, fair lending and consumer protection requirements, as the framework itself notes: 'SI does not change the bank's underlying legal obligations'.

The line between the two 2026 bulletins is drawn in footnote 1 of the framework: the model risk management principles of Bulletin 2026-13 'apply to non-generative and non-agentic SI models', so the framework 'is more applicable to generative and agentic SI'. A machine-learning credit model is read against Bulletin 2026-13; a generative assistant or an agent that can act is what examiners assess with Bulletin 2026-54.

Two items are still open. The interagency request for information on model risk management and banks' use of AI, which the OCC, Federal Reserve and FDIC said in Bulletin 2026-13 they would issue 'in the near future', had not been published when Bulletin 2026-54 appeared on November 4, 2026. And the proposed third-party guidance is out for comment until December 16, 2026, after Bulletin 2026-57 of November 13, 2026 extended the original November 16 deadline.

What responsible AI practices are expected of large banks as they innovate with AI?

The responsible AI practices expected of large banks are now specific enough to list. In the United States the April 2026 interagency model risk guidance (OCC Bulletin 2026-13, SR 26-2, FIL-15-2026) is the operative standard for banking organizations above roughly $30 billion in assets: an inventory of models, sound development and testing, independent validation with outcomes analysis, effective challenge, ongoing monitoring and board-level governance, all scaled to materiality. Generative and agentic AI are deliberately left outside it; since November 4, 2026 OCC examiners assess them under the Super Intelligence supervisory framework (Bulletin 2026-54) — sixteen components, six pillars and a four-tier, consequence-based classification that sets no expectations for banks — alongside third-party risk management (SR 23-4), operational resilience and cybersecurity expectations, and consumer-protection law, with the OCC's Spring 2026 Risk Perspective describing the observed large-bank practice as use limited to defined cases, guardrails, and human-in-the-loop accountability. Internationally the FSB's twelve sound practices for responsible AI adoption, finalised October 15, 2026 after a June 2026 consultation, give the fullest statement: board-set strategic direction and oversight, clear accountability, AI inside the enterprise risk framework, organisational skills, materiality-based risk assessment, model selection fit for purpose, data governance, explainability, performance monitoring, human oversight that tightens as autonomy grows, cyber and ICT controls, and third-party AI risk management. Acting Comptroller Hood's April 2025 formulation — robust risk management for AI consistent with conventional model risk management, covering bias, security vulnerabilities and explainability — the ECB's three governance expectations, the PRA's SS1/23 principles and New York DFS's AI-cyber and frontier-model letters all describe the same programme in different words.

RuleAuthorityWhat it requiresApplies
OCC Bulletin 2026-13 / SR 26-2 / FIL-15-2026OCCRisk-based, materiality-driven model risk management — sound development and testing, independent validation with outcomes analysis, effective challenge, board and senior-management governance — most relevant above $30 billion in assets. Generative and agentic AI are outside its scope and must be managed through broader governance programs.Since Apr 17, 2026
OCC Bulletin 2026-54 (Super Intelligence: Supervisory Framework)OCCThe OCC examiner's lens on generative and agentic AI: a documented purpose and permitted use, consequence-based tiering, human-in-the-loop or human-on-the-loop controls with real authority, MRM extensions for non-deterministic systems, data and privacy governance, vendor and provider-agent visibility, tested containment and kill switches, adversarial and hallucination testing, reconstructable records, independent audit and board reporting. Sets no supervisory expectations for banks.Issued Nov 4, 2026
Acting Comptroller Hood, Responsible AI SymposiumOCC'Robust risk management practices for AI applications, consistent with conventional model risk management' — covering bias, security vulnerabilities and explainability — and existing OCC and interagency guidance applied as the AI governance baseline.Since Apr 2025
OCC Semiannual Risk Perspective, Spring 2026OCCReports the observed large-bank practice for generative and agentic AI: a measured approach, use limited to specific cases, guardrails and human-in-the-loop accountability — and warns that lack of explainability makes governance 'essential' before use expands to material financial decisions.May 2026
SR 23-4 / OCC Bulletin 2023-17 / FIL-29-2023Federal ReserveVendor-supplied AI — cloud-hosted models, foundation-model access, embedded AI in platforms — goes through the full third-party lifecycle: due diligence, contracts with audit and data rights, ongoing monitoring, exit. The bank stays accountable for the outcome; supervisors said in May 2026 they are assessing how these expectations apply to vendor AI.In force
Vice Chair Bowman, FSOC AI roundtableFederal ReserveNo pre-emptive AI-specific rulemaking: existing frameworks, regularly reviewed, should accommodate AI. Confirms the generative/agentic carve-out from model risk guidance and frames frontier AI as dual-use in cybersecurity.May 2026
FSB 12 sound practices for responsible AI adoptionFSBOrganisation-wide: strategic direction and oversight, governance and accountability, AI in the risk framework, organisational adaptability. Lifecycle: materiality and risk assessment, model selection, data governance, explainability, performance management, human oversight (extra measures for highly autonomous agentic AI), cyber/ICT, third-party AI risk.Finalised Oct 15, 2026
NIST AI RMF 1.0 (AI 100-1)NISTVoluntary Govern–Map–Measure–Manage cycle and seven trustworthiness characteristics (valid and reliable, safe, secure and resilient, accountable and transparent, explainable, privacy-enhanced, fair). The common vocabulary US agencies and Treasury keep pointing to.Since Jan 2023
Treasury FS AI RMF and AI LexiconU.S. TreasuryAdapts the NIST AI RMF to financial-services operations, regulation and consumer protection; gives institutions tools to evaluate AI use cases across the lifecycle and a shared lexicon. Non-binding, scalable by size.Since Feb 19, 2026
NIST AI 600-1 (Generative AI Profile)NISTTwelve generative-AI risks — confabulation, information security (prompt injection, data poisoning, model extraction), data privacy, harmful bias, value-chain and component integration — with 200+ suggested actions mapped to RMF subcategories. The de facto control catalogue for the AI that model risk guidance excludes.Since Jul 2024
ECB: 'Technology is neutral, governance is not'ECBThree governance expectations for the 85%+ of large European banks using AI: clear accountability for AI outcomes, senior-management oversight matching AI's strategic importance, effective challenge from risk, compliance and internal audit — plus explainability, lifecycle governance and drift monitoring, data quality and third-party risk.Feb 2026
SSM supervisory priorities 2026–28ECBBanks 'shall have strategies that effectively reflect opportunities and risks' of new technologies; continued monitoring of general AI use and a targeted, in-depth approach to generative-AI applications, with DORA on-site campaigns on cyber and third-party risk.2026–28
PRA SS1/23UK (BoE / PRA / FCA)Five model-risk principles for all models informing business decisions regardless of technology, including vendor models, with AI-relevant tiering factors (unstructured data, explainability, bias) and a named Senior Management Function holder accountable for the framework.Since May 17, 2024
DFS Industry Letter on frontier AI modelsNY DFSFor New York-regulated institutions: shorter remediation timelines, mapped third-party dependencies, human review of AI-generated code before deployment, stronger logging and alerting, more frequent resilience testing — under 23 NYCRR Part 500.May 2026
FDIC Chairman Hill, oversight testimonyFDICThe FDIC's framing of the same guidance: the 2026 model-risk revision 'supports the use of innovative technology' and the BSA programme proposal encourages AI to detect illicit finance — responsible innovation is the supervisory expectation, not a bar to adoption.Jun 4, 2026
DFS Industry Letter on AI cybersecurity risksNY DFSAI threats — deepfake social engineering, AI-enhanced attacks, exposure of AI training data, AI supply-chain dependencies — must be in the risk assessment, training, MFA design and vendor diligence of every New York-regulated bank.Oct 16, 2024
Treasury report on AI-specific cybersecurity risks in financial servicesU.S. TreasuryThe practices Treasury asked large institutions for: AI in risk-management frameworks, data-supply-chain mapping, a common AI lexicon, information sharing on AI threats, and closing the capability gap with smaller institutions.Mar 2024
FSB Chair's letter to the G20 on frontier AI modelsFSBBoards should expect frontier models' autonomy and threat capability to change the speed and scale of cyber attacks; robust response and recovery capabilities and critical-provider resilience are the asks of firms.Aug 31, 2026
NIST CAISI request for information on AI agent securityNISTThe federal workstream on constraining and monitoring agent access, indirect prompt injection and misaligned behaviour — the technical content behind 'human-in-the-loop accountability' for agents.Jan 8, 2026
Basel Committee principles for third-party risk managementBasel CommitteeInternational principles for banks' reliance on third parties, including AI and cloud providers: criticality assessment, due diligence, contractual rights, monitoring, concentration and exit — the basis for the vendor half of responsible AI.2025
CFPB issue spotlight on chatbotsCFPBEvery top-10 US bank runs a chatbot; institutions remain responsible for accurate answers, dispute recognition and access to a human 'regardless of the processes or technologies used' — UDAAP, Regulation E and Z duties attach to the AI front end.Since Jun 2023
SEC Examination Priorities FY2026SECFor bank-affiliated broker-dealers and advisers: accurate representations of AI capabilities, and policies and procedures to monitor and supervise AI used in fraud prevention, back-office operations, AML and trading.FY2026
Bank of England / FCA 2026 AI surveyUK (BoE / PRA / FCA)The supervisors' own adoption data: 86% of 196 responding UK firms use AI, generative AI is 56% of reported AI systems, 80% have AI agents deployed and only 2% report agents with broad autonomy — the rest recommend while humans decide and execute.Published Oct 26, 2026
EU AI Act, Annex III 5(b) and Arts. 9–15, 26EU AI ActFor large banks with EU operations: creditworthiness scoring of natural persons is high-risk — risk management, data governance, logging, human oversight and, for deployers, monitoring, six-month log retention and a fundamental-rights impact assessment.From Dec 2, 2027

'Large' has a number in the United States. The April 2026 interagency model risk guidance says it is most relevant to banking organizations with more than $30 billion in total assets, and everything above that line is examined against it: an inventory of models, development evidence, independent validation with outcomes analysis, effective challenge, and a board that owns the framework. Below the line the same disciplines apply proportionately. The guidance replaced SR 11-7 after fifteen years and narrowed the definition of a model to complex quantitative methods — deterministic rules and simple arithmetic are out — which is why the OCC rescinded Bulletins 2011-12, 1997-24 and 2021-19 and the Comptroller's Handbook booklet on the same day.

What large banks actually do is now on the record from supervisors, not vendors. The OCC's Spring 2026 Risk Perspective describes generative and agentic AI use as 'primarily productivity and customer experience enhancement tools', deployed 'with guardrails and human-in-the-loop accountability' and limited to specific cases, while noting banks may expand to 'material financial decisions'. The ECB's Pedro Machado reported in February 2026 that more than 85% of large European banks use AI, with generative and agentic tools accelerating in IT operations, legal and document analysis and front-line support. The EBA's September 2025 survey put EU adoption at 92%, with 55% of banks already using general-purpose or agentic AI in consumer-facing processes. The CFPB found in 2023 that all of the top ten US commercial banks ran chatbots. The UK's 2024 survey found 75% adoption with only 2% of use cases fully autonomous; the 2026 edition, published 26 October 2026 and the first to ask about agentic AI, found 86% of 196 responding firms using AI and 80% with AI agents deployed, but only 2% deploying agents with broad autonomy.

The gap every large bank's AI governance has to bridge is explicit: the 2026 model risk guidance says generative and agentic AI 'are not within the scope of this guidance', and the agencies have promised an interagency request for information on AI and model risk 'in the near future'. On November 4, 2026 the OCC filled its side of the gap from the examiner's direction: Bulletin 2026-54, the Super Intelligence supervisory framework, tells OCC examiners what to look at in a generative or agentic use case — purpose boundaries, human oversight that is 'more than a nominal click', kill switches and circuit breakers, provider-agent limits, hallucination and adversarial testing, reconstructable records — scaled to four consequence-based tiers, while stating that it creates no supervisory expectations for banks. Until the RFI lands, four further frameworks fill the rest of the space. Third-party guidance (SR 23-4) governs the vendor models and cloud-hosted foundation models most banks actually consume — Vice Chair Bowman said in May 2026 supervisors are assessing exactly how it applies to vendor AI. NIST AI 600-1 supplies the control catalogue for generative-AI-specific risks such as confabulation, prompt injection and value-chain integration. NIST's CAISI has an open workstream on AI agent security — indirect prompt injection, misaligned behaviour, constraining and monitoring agent access — from its January 2026 RFI. And the FSB's Sound Practice 10 sets the international expectation that human oversight scales up, not down, as agents become more autonomous.

Direction of travel matters as much as the current rules, and it runs in two directions at once. On adoption, the posture is pro-innovation: Executive Order 14179 and the July 2025 AI Action Plan frame Treasury's February 2026 FS AI RMF; FSOC's 2025 report made 'harnessing AI' a priority and created a standing AI Working Group; the four-roundtable FSOC AI Innovation Series closed in June 2026 with participants asking for regulatory clarity and harmonization; FDIC Chairman Hill told Congress the revised model risk guidance 'supports the use of innovative technology'; HM Treasury's July 2026 plan wants firms 'beyond isolated pilots'; and the PRA's April 2026 plan keeps regulation technology-agnostic. On cyber, the warnings escalate: DFS's May 2026 frontier-AI letter, the ECB's 'Dear CEO' letter of July 2026 requiring AI-cyber action plans by 31 October, the ESAs' July 2026 statement under DORA, and the FSB Chair's 31 August 2026 letter naming frontier models' 'increasingly sophisticated autonomy' as a stability concern. A large bank's responsible-AI programme has to satisfy both audiences with one set of controls.

The list above is what 'responsible practices' means when supervisors use the phrase, and it is worth being literal about it, because the phrase is now the examination frame. A large bank is expected to show a model inventory tiered by materiality; validation evidence with outcomes analysis for anything that meets the 2026 definition of a model; a documented governance path for generative and agentic tools that names which third-party, cyber and consumer controls apply; a vendor file for every foundation model and platform it consumes; a defined human-oversight point for each AI system that scales up with autonomy; explainability sufficient for the decision the system supports; an AI threat entry in its cyber risk assessment; and board minutes showing the strategic direction and appetite were set at the top. Those eight items map one-to-one onto the FSB's practices, the OCC's observed conduct, DFS's letters and the ECB's expectations.

In practice the examination question is rarely 'do you have a responsible AI policy'. It is whether the bank can show, for any given AI system, which tier it sits in, who owns it, what it was validated against, what a human can override, what the vendor contract lets the bank see, and what happens when it is wrong. The documents above are the sources of each of those questions.

WHAT THIS MEANS IN PRACTICE

  • Tier every AI system by materiality before choosing the control set. Above $30 billion, anything that meets the 2026 definition of a model gets full validation; generative and agentic tools get a documented governance path that names which of SR 23-4, NIST 600-1 and the operational-resilience controls apply.
  • Write the generative/agentic gap into policy rather than leaving it implicit: the guidance itself says these tools are managed through 'broader risk management and governance' — examiners will ask to see that program, and since November 2026 OCC examiners carry Bulletin 2026-54's sixteen components and four tiers as the question list for it.
  • Treat vendor AI as a third-party relationship first and a model second. Due diligence, contract audit rights, data-use terms and exit plans are what supervisors are currently assessing for AI vendors.
  • Make human-in-the-loop measurable. The observed large-bank practice supervisors praise is not a checkbox; it is a defined reviewer with authority to override, evidence they actually review, and a threshold at which automation stops.
  • Fund explainability as a control, not a research project. Hood in 2025, the Risk Perspective in 2026 and the ECB in 2026 all name explainability as the governance issue that limits where AI can be used.
  • Give internal audit and compliance an effective-challenge role over AI — the ECB's third expectation and the FSB's second practice — rather than housing all AI oversight in technology.
  • Map the cyber warnings to remediation timelines now: human review of AI-generated code, faster patching, dependency maps and logging are what DFS, the ECB and the ESAs asked for in the same eight weeks of 2026.
  • Keep the eight evidence items above as a standing pack — inventory, validation, generative/agentic governance path, vendor files, human-oversight points, explainability, AI cyber risk entry, board minutes — so any supervisor's 'responsible AI' question can be answered with documents rather than a policy statement.
  • Publish the practices the bank actually follows in its own words; supervisors and the FSB consultation both draw on what large banks disclose, and the disclosure is now part of how peers are assessed.
  • Watch for the interagency RFI on AI and model risk management: it is the first federal document that will speak to generative and agentic AI directly, and comment letters from large banks will shape it.

What has the OCC actually published on AI?

DateDocumentStatus
Nov 4, 2026OCC Bulletin 2026-54 — Super Intelligence: Supervisory FrameworkFinal
May 7, 2026OCC Semiannual Risk Perspective, Spring 2026 — Semiannual Risk Perspective from the National Risk Committee, Spring 2026Final
Apr 17, 2026OCC Bulletin 2026-13 — Model Risk Management: Revised GuidanceIn force
Apr 29, 2025Acting Comptroller Hood, 'AI in Financial Services' (Apr 2025) — Remarks by Acting Comptroller Rodney E. Hood at the National Fair Housing Alliance's Responsible AI Symposium: 'AI in Financial Services'Final
Jun 6, 2023OCC Bulletin 2023-17 — Third-Party Relationships: Interagency Guidance on Risk ManagementIn force
Apr 9, 2021OCC Bulletin 2021-19 — Bank Secrecy Act/Anti-Money Laundering: Interagency Statement on Model Risk Management for Bank Systems Supporting BSA/AML Compliance and Request for InformationSuperseded
Mar 31, 20212021 Interagency AI RFI (OCC Bulletin 2021-17) — Request for Information and Comment on Financial Institutions' Use of Artificial Intelligence, Including Machine LearningFinal
Apr 4, 2011OCC Bulletin 2011-12 — Sound Practices for Model Risk Management: Supervisory Guidance on Model Risk ManagementSuperseded
May 20, 1997OCC Bulletin 1997-24 — Credit Scoring Models: Examination GuidanceSuperseded
DateTypeDocument / event
Nov 4, 2026FrameworkOCC Bulletin 2026-54 — Super Intelligence: Supervisory Framework. OCC Bulletin 2026-54, issued November 4, 2026, transmits the OCC's 'Super Intelligence: Supervisory Framework', a 20-page resource for OCC examiners evaluating banks' use of Super Intelligence (SI) — the OCC's term, under Executive Order 14434, for the technologies previously called artificial intelligence — with a focus on generative and agentic systems. source ↗
May 7, 2026ReportOCC Semiannual Risk Perspective, Spring 2026 — Semiannual Risk Perspective from the National Risk Committee, Spring 2026. The OCC's Spring 2026 Semiannual Risk Perspective, released May 7, 2026, says artificial intelligence is 'significantly transforming the cyber threat landscape' — lowering the barrier to entry for attackers and increasing the 'speed, scale, and sophistication' of cyberattacks and fraud — while also giving banks new defensive capabilities. source ↗
Apr 17, 2026GuidanceOCC Bulletin 2026-13 — Model Risk Management: Revised Guidance. On April 17, 2026 the OCC, Federal Reserve, and FDIC issued revised interagency Model Risk Management guidance (OCC Bulletin 2026-13; Fed SR 26-2; FDIC FIL-15-2026), replacing the 2011 framework that had governed bank models for 15 years. source ↗
Apr 29, 2025SpeechActing Comptroller Hood, 'AI in Financial Services' (Apr 2025) — Remarks by Acting Comptroller Rodney E. Hood at the National Fair Housing Alliance's Responsible AI Symposium: 'AI in Financial Services'. On April 29, 2025 Acting Comptroller Rodney E. source ↗
Jun 6, 2023GuidanceOCC Bulletin 2023-17 — Third-Party Relationships: Interagency Guidance on Risk Management. OCC Bulletin 2023-17, issued June 6, 2023, transmits the interagency Guidance on Third-Party Relationships: Risk Management from the OCC, Federal Reserve, and FDIC. source ↗
Apr 9, 2021GuidanceOCC Bulletin 2021-19 — Bank Secrecy Act/Anti-Money Laundering: Interagency Statement on Model Risk Management for Bank Systems Supporting BSA/AML Compliance and Request for Information. OCC Bulletin 2021-19, issued April 9, 2021, transmitted the interagency Statement on Model Risk Management for Bank Systems Supporting BSA/AML Compliance, issued by the OCC, Federal Reserve, FDIC, and NCUA in consultation with FinCEN, together with a request for information (comments due June 11, 2021). source ↗
Mar 31, 2021Consultation2021 Interagency AI RFI (OCC Bulletin 2021-17) — Request for Information and Comment on Financial Institutions' Use of Artificial Intelligence, Including Machine Learning. On March 31, 2021 the OCC, Federal Reserve, FDIC, CFPB, and NCUA jointly published a Request for Information on financial institutions' use of artificial intelligence, including machine learning — the first coordinated federal inquiry into bank AI. source ↗
Apr 4, 2011GuidanceOCC Bulletin 2011-12 — Sound Practices for Model Risk Management: Supervisory Guidance on Model Risk Management. OCC Bulletin 2011-12, 'Sound Practices for Model Risk Management', issued April 4, 2011, transmitted the interagency Supervisory Guidance on Model Risk Management (the Federal Reserve's SR 11-7). source ↗
May 20, 1997GuidanceOCC Bulletin 1997-24 — Credit Scoring Models: Examination Guidance. OCC Bulletin 1997-24, 'Credit Scoring Models: Examination Guidance,' issued May 20, 1997, was the OCC's first supervisory statement on algorithmic credit decisions. source ↗

Which of the 120 largest US banks answer to the OCC on AI?

49 of the 120 bank pages on this site name the OCC among the authorities their AI programme answers to. Each page lists the documents that apply and why.

Which fintechs does the OCC supervise or license?

9 of the 56 fintech and neobank pages on this site name the OCC as a direct supervisor or licensing authority. Each page sets out the AI programme and the rules that apply to it.

  • The interagency request for information on model risk management and banks' use of AI (including generative and agentic AI) that the OCC, Fed, and FDIC promised 'in the near future' in Bulletin 2026-13 and repeated in the Spring 2026 Semiannual Risk Perspective — still not published as of November 2, 2026
  • The Fall 2026 Semiannual Risk Perspective (normally published Nov/Dec) and its AI section — not yet out as of November 2, 2026
  • The OCC's stated review of supervisory expectations, guidance, and regulations to 'right-size' AI expectations for community banks that rely on third-party technology (Spring 2026 Semiannual Risk Perspective)
  • Revisions of the Super Intelligence supervisory framework (Bulletin 2026-54, November 4, 2026), which the OCC says it will update regularly as technology, use cases and supervisory understanding mature — and whether the Federal Reserve and FDIC issue an equivalent examiner document for generative and agentic AI
  • Whether the final interagency third-party guidance absorbs the SI-specific vendor considerations of Bulletin 2026-54 (provider-agent visibility, SI incident vocabularies, liability for model errors, notice of new SI features)
  • The OCC's September 11, 2026 proposed interagency Third-Party Risk Management Guidance (Bulletin 2026-46), issued with the Fed, FDIC and NCUA to replace the 2023 third-party framework banks use for vendor and foundation-model AI; comments due December 16, 2026, extended from November 16 by Bulletin 2026-57 (November 13, 2026)
  • Interaction between the deregulatory 2025–26 posture and consumer-protection enforcement on AI lending

Does SR 11-7 / OCC 2011-12 still apply to AI models?

No — as of April 17, 2026 the 2011 interagency model risk guidance was superseded by revised guidance from the OCC, Federal Reserve, and FDIC (OCC Bulletin 2026-13). Traditional and machine-learning models fall under the revised framework; generative and agentic AI are explicitly excluded and are instead governed through banks' broader risk-management programs.

Why did the 2026 guidance exclude generative AI?

The agencies concluded generative and agentic AI are 'novel and rapidly evolving' — too fast-moving for prescriptive model-validation requirements. Banks are expected to apply enterprise risk management and governance controls instead. On November 4, 2026 the OCC filled its side of the gap with Bulletin 2026-54, the Super Intelligence supervisory framework, which tells examiners how to assess those systems without creating new expectations for banks; the interagency request for information promised in April 2026 had still not appeared.

What is the OCC Super Intelligence supervisory framework?

OCC Bulletin 2026-54, issued November 4, 2026, transmits a 20-page framework for OCC examiners evaluating banks' use of 'Super Intelligence' (SI) — the OCC's term, under Executive Order 14434, for AI — with a focus on generative and agentic systems. It has sixteen components in six pillars (purpose, risk, control, oversight, resilience, assurance), an illustrative four-tier classification based on the consequences of failure, reversibility and autonomy rather than technical sophistication, and a lexicon. The bulletin says the framework does not establish supervisory expectations for banks, sets no enforceable standards, and will be updated regularly. Model risk management for non-generative, non-agentic models stays under Bulletin 2026-13.

Is there any binding US federal AI regulation for banks?

There is no AI-specific federal statute for banks. AI use is regulated through existing law — safety and soundness, fair lending (ECOA), UDAP/UDAAP — and supervisory guidance like the 2026 model risk framework. This contrasts with the EU, where the AI Act imposes binding AI-specific obligations.

Which OCC bulletins did Bulletin 2026-13 rescind?

Four issuances: OCC Bulletin 2011-12 (Supervisory Guidance on Model Risk Management), OCC Bulletin 1997-24 (Credit Scoring Models: Examination Guidance), OCC Bulletin 2021-19 (the interagency statement on model risk management for BSA/AML compliance), and the 'Model Risk Management' booklet of the Comptroller's Handbook. The rescinded bulletins remain readable in the OCC's rescinded-bulletins archive.

How does the OCC's approach compare?

Follow every move these regulators make

when one of these regulators moves, the next morning's brief says so · six sourced stories · 7 am PT · free

plus every regulation, bank and agent page update, the morning after · leave any morning