SEC filings · computed from SEC EDGAR filings dataset; nothing on this page is written by a model. All ranked answers: /banks/leaders
On the public record as of Oct 31, 2026, 66 of the 141 large banks profiled have a verbatim filing statement classified under model risk. Ranked by quotes (filings), the top 5 are BOK Financial (12 filing quotes), State Street (9 filing quotes), Truist (6 filing quotes), Wells Fargo (6 filing quotes) and Huntington (5 filing quotes). BOK Financial's strongest evidence: BOK Financial says its Model Risk Management unit enforces model governance and can require remediation or restrict model usage. (Form 10-Q, MD&A, filed Aug 4, 2026).
141 banks profiled (120 US, 21 non-US) · 66 with matching evidence · top 40 shown · ranked answer, not a comparison
Which of the large banks discuss AI model risk in their SEC filings?
| # | Bank | Quotes (filings) | Latest statement (extractor's one-line claim) | Status | Date | On this site | Sources |
|---|---|---|---|---|---|---|---|
| 1 | BOK Financial | 12 | BOK Financial says its Model Risk Management unit enforces model governance and can require remediation or restrict model usage. (Form 10-Q, MD&A, filed Aug 4, 2026) | 6 filings | Aug 4, 2026 | BOK Financial · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 2 | State Street | 9 | State Street says model results influence its business decisions and model failure could harm its financial performance. (Form 10-Q, MD&A, filed Jul 30, 2026) | 8 filings | Jul 30, 2026 | State Street · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 3 | Truist | 6 | Truist says its Model Risk Oversight unit independently validates all decision models, including trading market risk models. (Form 10-Q, MD&A, filed May 1, 2026) | 5 filings | May 1, 2026 | Truist · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 4 | Wells Fargo | 6 | Wells Fargo notes that complex, opaque AI models make it hard to understand why they produce particular outputs. (Form 10-K, risk factors, filed Feb 24, 2026) | 5 filings | Feb 24, 2026 | Wells Fargo · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 5 | Huntington | 5 | Huntington says it continuously monitors model performance and may update its statistically based models as enhancements are identified. (Form 10-Q, MD&A, filed Apr 30, 2026) | 5 filings | Apr 30, 2026 | Huntington · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 6 | Morgan Stanley | 5 | Morgan Stanley defines model risk as adverse consequences from incorrect or misused model outputs, including financial loss and reputational damage. (Form 10-Q, MD&A, filed May 5, 2026) | 4 filings | May 5, 2026 | Morgan Stanley · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 7 | JPMorgan | 5 | JPMorgan says its market risk models are inherently imprecise and may be less reliable during sudden or severe market shifts. (Form 10-Q, other, filed May 1, 2026) | 4 filings | May 1, 2026 | JPMorgan · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 8 | Regions | 4 | Regions' Risk Committee monitors model performance, including generative AI, among other key risks. (Form DEF 14A, proxy, filed Mar 23, 2026) | 4 filings | Mar 23, 2026 | Regions · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 9 | Schwab | 4 | Schwab says it manages AI model risk through policies, controls, and cross-functional review before any AI is used. (Form 10-K, MD&A, filed Feb 25, 2026) | 3 filings | Feb 25, 2026 | Schwab · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 10 | KeyBank | 4 | KeyCorp says some of its models use AI or machine learning, adding complexity such as data needs, algorithmic bias and interpretability challenges. (Form 10-K, risk factors, filed Feb 23, 2026) | 3 filings | Feb 23, 2026 | KeyBank · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 11 | HSBC USA | 4 | HSBC USA defines model risk as the potential for adverse consequences from model errors or inappropriate use of modelled outputs in business decisions. (Form 10-Q, MD&A, filed Apr 29, 2025) | 3 filings | Apr 29, 2025 | HSBC USA · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 12 | Atlantic Union | 4 | Atlantic Union says some of its models use AI or machine learning, which carry unique complexities such as training data needs, bias potential, and interpretability… (Form 10-K, risk factors, filed Feb 26, 2026) | 2 filings | Feb 26, 2026 | Atlantic Union · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 13 | PNC | 4 | PNC cautions that automated AI tools depend on data quality and that humans may better detect anomalous data or results. (Form 10-K, risk factors, filed Feb 20, 2026) | 2 filings | Feb 20, 2026 | PNC · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 14 | Citi | 3 | Citi's credit loss reserve qualitative adjustments account for risks from AI to vulnerable sectors that the quantitative model does not fully capture. (Form 10-Q, MD&A, filed May 7, 2026) | 3 filings | May 7, 2026 | Citi · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 15 | Zions | 3 | Zions says deposit models are updated when behavior diverges and all model changes are independently reviewed by its Model Risk Management function. (Form 10-Q, MD&A, filed May 7, 2026) | 3 filings | May 7, 2026 | Zions · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 16 | Western Alliance | 3 | A Model Risk Management Committee oversees the company's Model Risk Management Framework and reports model risk to the ERMC. (Form DEF 14A, proxy, filed Apr 22, 2026) | 3 filings | Apr 22, 2026 | Western Alliance · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 17 | Ally | 3 | Ally says responsible AI practices are built into design, approval and operational controls, with continuous testing and independent model validation. (Form DEF 14A, proxy, filed Mar 18, 2026) | 3 filings | Mar 18, 2026 | Ally · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 18 | Amex | 3 | Amex warns that its and its partners' AI and ML use carries risks such as biased or inaccurate results and hallucinated outputs. (Form 10-K, risk factors, filed Feb 6, 2026) | 3 filings | Feb 6, 2026 | Amex · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 19 | Raymond James | 3 | Raymond James says model errors or misuse could cause significant financial loss and inaccurate reporting, with models used for valuation, credit allowances and stress… (Form 10-K, MD&A, filed Nov 25, 2025) | 3 filings | Nov 25, 2025 | Raymond James · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 20 | Eastern Bank | 3 | Eastern uses a quantitative loan rating model for CECL-style lifetime loss estimates, with qualitative overlays to address model risk. (Form 10-Q, MD&A, filed Aug 8, 2025) | 3 filings | Aug 8, 2025 | Eastern Bank · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 21 | First Hawaiian | 3 | First Hawaiian says it closed key supervisory recommendations on model risk and data governance, strengthening regulatory credibility. (Form DEF 14A, proxy, filed Mar 12, 2026) | 2 filings | Mar 12, 2026 | First Hawaiian · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 22 | Renasant | 3 | Renasant says it relies on third-party AI models, so it depends partly on how those third parties develop and train them. (Form 10-K, risk factors, filed Mar 2, 2026) | 2 filings | Mar 2, 2026 | Renasant · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 23 | NBT | 3 | NBT says it uses AI across platforms, including third-party systems, and evaluates AI tools through its SDLC with model validation and risk controls. (Form 10-K, business, filed Feb 27, 2026) | 2 filings | Feb 27, 2026 | NBT · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 24 | Bank of Hope | 3 | Hope Bancorp's Model Risk Management Department validates models, evaluating reasonableness and sensitivity of significant judgments and assumptions. (Form 10-K, other, filed Feb 25, 2026) | 2 filings | Feb 25, 2026 | Bank of Hope · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 25 | Sallie Mae | 3 | SLM warns AI models could leak confidential information or infringe IP, and their opacity makes risks hard to assess and mitigate. (Form 10-K, risk factors, filed Feb 19, 2026) | 2 filings | Feb 19, 2026 | Sallie Mae · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 26 | Simmons Bank | 2 | Simmons says third-party AI models carry risks in how they are developed, trained and deployed, including unauthorized training data and limited visibility into… (Form 10-Q, risk factors, filed Aug 5, 2026) | 2 filings | Aug 5, 2026 | Simmons Bank · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 27 | FNB | 2 | FNB says its AI Policy and Model Risk Management Policy target prudent AI deployment with focus on transparency, explainability and bias mitigation. (Form DEF 14A, proxy, filed Mar 25, 2026) | 2 filings | Mar 25, 2026 | FNB · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 28 | Santander US | 2 | Santander warns AI models may be flawed, biased or trained on poor data, with errors that may not be easily detectable. (Form 10-K, risk factors, filed Mar 2, 2026) | 2 filings | Mar 2, 2026 | Santander US · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 29 | Banc of California | 2 | Banc of California warns third-party AI systems may produce inaccurate or biased outputs or leak confidential information, and its reliance on vendors limits its… (Form 10-K, risk factors, filed Feb 27, 2026) | 2 filings | Feb 27, 2026 | Banc of California · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 30 | Rockland Trust | 2 | Independent Bank warns generative AI models may produce incorrect, biased, or harmful output and leak confidential information. (Form 10-K, risk factors, filed Feb 27, 2026) | 2 filings | Feb 27, 2026 | Rockland Trust · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 31 | First Horizon | 2 | First Horizon assigns trading risk management personnel primary responsibility for model risk management of its VaR model. (Form 10-K, MD&A, filed Feb 26, 2026) | 2 filings | Feb 26, 2026 | First Horizon · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 32 | Banner Bank | 2 | Banner warns that flawed algorithms or biased data could cause compliance violations or discriminatory lending outcomes. (Form 10-K, risk factors, filed Feb 25, 2026) | 2 filings | Feb 25, 2026 | Banner Bank · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 33 | Northern Trust | 2 | Northern Trust applies its model risk framework, including independent validation and monitoring, to AI and machine learning models. (Form 10-K, MD&A, filed Feb 24, 2026) | 2 filings | Feb 24, 2026 | Northern Trust · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 34 | U.S. Bank | 2 | U.S. Bank says its AI and machine learning models carry complexities including data needs, algorithmic bias and explainability challenges that can reduce accuracy. (Form 10-K, risk factors, filed Feb 23, 2026) | 2 filings | Feb 23, 2026 | U.S. Bank · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 35 | Capital One | 2 | Capital One says it relies on quantitative models and some AI for risk management, estimates, forecasts and capital compliance. (Form 10-K, risk factors, filed Feb 19, 2026) | 2 filings | Feb 19, 2026 | Capital One · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 36 | SoFi | 2 | SoFi says its own AI use is limited but its systems and partners' systems increasingly rely on complex AI that may contain undetectable errors. (Form 10-K, risk factors, filed Feb 17, 2026) | 2 filings | Feb 17, 2026 | SoFi · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 37 | WaFd Bank | 2 | WaFd warns generative or agentic AI models may produce incorrect, biased, privacy-breaching, infringing or harmful outputs or actions. (Form 10-K, business, filed Nov 18, 2025) | 2 filings | Nov 18, 2025 | WaFd Bank · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 38 | Axos Bank | 2 | Axos warns generative AI models may produce incorrect, biased, confidential-leaking or infringing output. (Form 10-K, risk factors, filed Aug 21, 2025) | 2 filings | Aug 21, 2025 | Axos Bank · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 39 | BNY | 2 | BNY's risk committee reviews model risk among its significant risk exposures overseen at Board level. (Form DEF 14A, proxy, filed Mar 5, 2025) | 2 filings | Mar 5, 2025 | BNY · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
| 40 | BofA | 2 | BofA warns that its models have limitations, including emerging risks from AI-reliant applications. (Form 10-K, risk factors, filed Feb 25, 2025) | 2 filings | Feb 25, 2025 | BofA · in their words → | SEC EDGAR ↗ · SEC EDGAR ↗ |
Each row quotes one record from SEC EDGAR filings dataset; the “On this site” link opens the bank's own page or sub-page with every source. Banks absent from the table have nothing matching on the record this site holds.
How is this ranked?
Data: the SEC filings dataset (data/sec/quotes.jsonl): verbatim AI statements from 10-K, 10-Q, 8-K and DEF 14A filings of the banks' holding companies, each with its EDGAR link, filing date, section and topic.
Rule: Quotes the extractor classified as model risk (model validation, model error, reliance on models). Score = such quotes; ties by distinct filings, then the newest filing.
Counts: 66 of 141 profiled banks qualify; the table shows the top 40; 80 distinct sources cited in the table. Data checked 2026-10-31; ranking template updated 2026-10-30.
What are the caveats?
The dataset holds the AI passages the extractor found in filings since January 2024; a bank with few quotes may say more in passages not captured. Quotes are verbatim; the one-line claim is the extractor's summary and links the EDGAR text.
Banks without a US SEC filer (most non-US groups, private and mutual banks) have no quotes and are absent.
Banks absent from the table have nothing matching on the record this site holds, which is not evidence that they do nothing. Each row links the bank's own page, where every entry carries its sources.
Which bank ranks first?
BOK Financial Corporation leads with 12 filing quotes: BOK Financial says its Model Risk Management unit enforces model governance and can require remediation or restrict model usage. (Form 10-Q, MD&A, filed Aug 4, 2026). State Street is second with 9.
How many large banks discuss AI model risk in SEC filings?
66 of the 141 profiled banks (120 US, 21 non-US groups); the table shows the top 40. Among those listed, 40 are US banks and 0 non-US groups.
What counts, and what does not?
Quotes the extractor classified as model risk (model validation, model error, reliance on models). Score = such quotes; ties by distinct filings, then the newest filing. Source: the SEC filings dataset (data/sec/quotes.jsonl): verbatim AI statements from 10-K, 10-Q, 8-K and DEF 14A filings of the banks' holding companies, each with its EDGAR link, filing date, section and topic.
Is this a measure of capability?
No. Every row is a sourced fact on the public record: what the bank, a regulator, a filing or an independent outlet has put on the record. The page does not rate how well anything works.
When was this checked, and how often does it update?
The data behind the table was last checked on 2026-10-31 (SEC EDGAR filings dataset). The page recomputes whenever that dataset changes and is republished with the site; it is never edited by hand.
When any of these banks moves on AI, you'll read it the next morning.
when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning