Can bank employees use ChatGPT? What GLBA, SAR secrecy, and exam rules say

The Morse teamPublished November 8, 202610 min read

Yes, bank employees can use ChatGPT for work with no customer, SAR, or exam information in it, but anything carrying those belongs only in an AI tool the bank has approved, as the codes of conduct at Bank of America and JPMorgan Chase now say. As of November 2026, a customer's name next to an account number is "sensitive customer information" under the federal banking agencies' security guidelines, a suspicious activity report (SAR) can't be disclosed outside the channels FinCEN's rule allows, and an exam report stays the regulator's property, so none of the three belongs in a personal ChatGPT account.

This isn't legal advice. It covers the rules that travel with bank information into a chatbot. On November 8, 2026, we read the Federal Reserve's SR 26-2 and the OCC's matching bulletin, the Interagency Guidelines Establishing Information Security Standards, Regulation P, the 2023 interagency guidance on third-party relationships, FinCEN's SAR confidentiality rule and the FFIEC's BSA/AML manual, the OCC's and the Fed's rules on confidential supervisory information, the Federal Register's list of OCC documents, the CFPB's chatbot spotlight, the codes of conduct of Bank of America and JPMorgan Chase, and CB Financial Services' May 2026 filing with the SEC. Bank of America deserves credit. Its code names ChatGPT and Google Gemini outright, so its employees don't have to guess.

Is it against bank policy to use ChatGPT at work?

At JPMorgan Chase and Bank of America, for anything about the bank's business, yes. Bank of America's 2026 Code of Conduct tells employees not to use "non-company-approved applications, third-party websites or devices for bank-specific processes, applications or data," and its examples include "generative AI chatbots like ChatGPT or Google Gemini" and AI search engines like "Perplexity AI." Its AI section says to use "Bank-approved tools only," and warns that unauthorized use of AI tools "may result in disciplinary action up to and including termination of employment."

The same code counts "generative AI tools" as written electronic communications, the category it requires employees to send only through "company-approved communication devices, approved applications and approved connecting services." A prompt you type about bank business falls under the same rule as a text message about it.

JPMorgan Chase's Code of Conduct, dated September 2026, doesn't name a chatbot. It says: "Do not use personal devices or any applications, tools, or services that are not approved by the firm to record, capture, store, or send information related to the firm's business." Its worked example has an employee using "an internal large language model" for a client meeting, and the answer puts the output on the employee: "You are responsible for the accuracy of information you create and share on behalf of the firm."

Smaller banks write their own policies. If yours hasn't published one, ask compliance before you assume a personal account is fine, and keep it off the bank's laptop, where IT can often see it anyway, as we cover in can my employer see my ChatGPT.

What happens when a bank employee puts customer data into an AI app?

One bank told the SEC. On May 11, 2026, CB Financial Services, the parent of Community Bank in Pennsylvania, filed an 8-K under Item 1.05, the item for material cybersecurity incidents (SEC). It said that on May 5 the bank "became aware of an internal incident involving the handling of certain non-public customer information using an unauthorized artificial intelligence-based software application."

There was no break-in. The filing says the incident "did not involve a disruption to the Bank's operations" or its core systems, but "due to the volume and sensitive nature of the non-public information at issue," the company judged it material on May 7. The information disclosed included "customer names, social security numbers and dates of birth," and the bank began customer notifications and talks with "relevant banking and financial regulators." As of November 8, 2026, the company's filings list no amendment naming the app or the number of customers.

Does GLBA let a bank send customer data to an AI company?

Only to one the bank has vetted and bound by contract. The Interagency Guidelines Establishing Information Security Standards (the OCC's copy, with matching FDIC and Fed versions) define a "service provider" as anyone that "maintains, processes, or otherwise is permitted access to customer information" through services to the bank. An AI company processing a loan file you pasted fits that definition. Section III.D then says the bank shall "exercise appropriate due diligence in selecting its service providers" and "require its service providers by contract to implement appropriate measures." A personal ChatGPT account has a contract with you, not with your bank.

The same appendix defines "sensitive customer information" as a customer's name, address, or phone number "in conjunction with" a Social Security number, driver's license number, account number, card number, or a password or PIN that opens the account. A loan application carries that pairing on its first page.

Regulation P adds a privacy rule. A bank may share nonpublic personal information with a service provider without offering customers an opt-out only if it has "a contractual agreement with the third party that prohibits the third party from disclosing or using the information other than to carry out the purposes for which you disclosed the information" (12 CFR 1016.13).

The agencies' 2023 third-party guidance closes the gap a free tool might seem to leave. It covers "any business arrangement between a banking organization and another entity, by contract or otherwise," and says "a third-party relationship may exist despite a lack of a contract or remuneration." It also says using a third party "does not diminish" the bank's own duty to comply. For the customer's side of the same question, see private AI for financial questions.

Can you put a SAR into ChatGPT?

No. FinCEN's rule for banks says "No bank, and no director, officer, employee, or agent of any bank, shall disclose a SAR or any information that would reveal the existence of a SAR" (31 CFR 1020.320(e)). The exceptions run to FinCEN, law enforcement, the regulators that examine for the Bank Secrecy Act, and sharing "within the bank's corporate organizational structure" for BSA purposes. An AI company is none of those.

The FFIEC BSA/AML manual adds that "the existence or even the non-existence of a SAR must be kept confidential." A prompt asking a chatbot to tighten a SAR narrative discloses the SAR. So does one asking whether a customer's activity is enough to file on, because it reveals that the bank is weighing one. The manual tells examiners who find an improper disclosure to make sure the bank has "notified the appropriate federal banking agency and FinCEN." The same section notes that a bank can't reveal a Right to Financial Privacy Act national security letter either.

Can you paste an exam report or an MRA into AI?

Not into a personal one. Exam findings are confidential supervisory information, and they belong to the agency. The OCC's rule says no bank or person "may disclose information covered by this subpart in any manner" without OCC approval, except to people "officially connected" with the bank, like its auditors and lawyers, or to a consultant under a written contract that states its "awareness of, and agreement to abide by" the ban on passing it on (12 CFR 4.37). Disclosing it otherwise "may be subject to the penalties provided in 18 U.S.C. 641."

The Fed's rule says its confidential supervisory information "remains the property of the Board," and no one who holds it may "use any such information for an unauthorized purpose or disclose" it beyond what the rule allows (12 CFR 261.20). A matter requiring attention pasted into a chatbot to draft the response plan is a disclosure under either rule.

Is there a federal rule on bank employees using generative AI?

Not a specific one, as of November 8, 2026. Guides written before April 2026 still point to SR 11-7, but on April 17, 2026, the Fed, the OCC, and the FDIC replaced it with SR 26-2, which "supersedes and replaces SR letter 11-7." The new guidance leaves generative AI out. Its attached text says: "Generative AI and agentic AI models are novel and rapidly evolving. As such, they are not within the scope of this guidance." It adds that a bank's own "risk management and governance practices should guide" controls for tools it doesn't cover.

The OCC's Bulletin 2026-13 says the agencies "plan to issue in the near future a request for information" on banks' use of AI, "including generative AI and agentic AI." As of November 8, 2026, the Federal Register lists no such request from the OCC since April. Until there is one, the rules above do the work: the security guidelines, Regulation P, the third-party guidance, SAR secrecy, and exam confidentiality.

The CFPB's 2023 spotlight on bank chatbots dealt with the bank's customer-facing bots, not staff use, and warned that a bot "may fail to protect their privacy and data." The CFPB now marks the page as archived content that "may be outdated."

What can a banker paste into AI?

We applied each rule's text, read on November 8, 2026, to the documents a banker handles most. "Personal AI" means a ChatGPT, Claude, or Gemini account your bank hasn't approved.

How Morse keeps a banker's AI questions private

Each rule above is about the readable copy of a prompt that lands with an AI company your bank never vetted, kept under that company's terms. Morse's servers don't keep your messages. In an E2EE chat, a verified enclave is the only thing that can read the message, and our servers pass along ciphertext. In any mode, the chat history lives in your browser rather than on our servers.

That doesn't make Morse a bank-approved vendor. Customer files, SARs, and exam findings stay in the tools your bank approved, and whether anything about the bank may go into Morse is your bank's call. Morse is for the questions that are yours: studying for a certification, reading a new rule, or thinking through your own career on your own device.

• E2EE, for questions about your own job that no one should read. A dispute with a manager, a move to another bank, or a question about your own rights. Your browser checks the secure enclave's hardware attestation, then encrypts the message so only that enclave can read it, and the reply is decrypted only in your browser. The enclaves are run by third-party providers, not by Morse, and the guarantee rests on the enclave hardware and attestation working as designed. E2EE runs open-weight models like GLM, Kimi, and DeepSeek, without web search.
• Private, for regulation study and general banking questions. New chats start in Private, on open-weight models like DeepSeek, Kimi, and GLM, plus xAI's Grok. Morse has zero data retention agreements with its inference, search, and other providers, so they keep nothing (what zero retention means).
• Anonymous, for GPT, Claude, or Gemini without the provider knowing it's you. Our proxy removes your account, email, and IP address and sends the message under Morse's own account. The provider sees the text and may keep it for a limited time to check for abuse, under its own policy, so keep to the last column of the table.

In every mode, Morse doesn't train AI on your content, sell your data, or show ads. Your history is encrypted in your browser and opens with two keys, one that never leaves your device and one our server gives you when you sign in. Because we don't store chats, we usually have very little to give authorities (privacy policy). Clearing the browser or deleting your account deletes your chats for good, and Morse can't recover them, so keep anything you need. For the same question in a neighboring field, see can insurance agents use ChatGPT.

Try a chat from the Morse home page without an account, or start free with no card. E2EE chats come with Pro, at $19 a month.