Can lawyers use ChatGPT? What the ethics opinions ask you to check first

The Morse teamPublished October 31, 20267 min read

Yes, lawyers can use ChatGPT, and as of October 2026 none of the ethics opinions we read forbids it. They limit what a lawyer types into it. ABA Model Rule 1.6(c) requires "reasonable efforts to prevent the inadvertent or unauthorized disclosure of" client information, ABA Formal Opinion 512 requires a client's informed consent before information about a representation goes into a "self-learning" tool, and consumer ChatGPT is set by default to use your chats for training.

This isn't legal advice or an ethics opinion; your own jurisdiction's rules decide what you may do.

On October 23, 2026, we read the comments to Model Rules 1.1, 1.6, and 5.3 as the ABA publishes them, D.C. Bar Opinion 388 and Texas Opinion 705 in full, Opinion 512's supervision and cloud computing sections, Florida's and California's opinions, and OpenAI's and Anthropic's business terms. All of them turn on where a client's information goes once you press send.

Is it ethical for lawyers to use ChatGPT?

Yes, if you understand the tool. Comment 8 to Model Rule 1.1 says a lawyer should keep up with "the benefits and risks associated with relevant technology" (ABA). Opinion 512, issued July 29, 2024, says lawyers "need not become GAI experts" but need "a reasonable understanding of the capabilities and limitations" of their tools (ABA Formal Opinion 512, pages 2 and 3).

Can lawyers put client information into ChatGPT?

Only after checking what the tool does with it, weighing "the sensitivity of the information" and "the likelihood of disclosure," as comment 18 to Rule 1.6 puts it (ABA). Opinion 512 worries most about tools that learn from what you type, where one client's facts can resurface in someone else's answer. It says that risk holds "even if the tool is used exclusively by lawyers at the same firm," since a colleague behind an ethical wall could get those facts back (page 6). Florida's Opinion 24-1, dated January 19, 2024, tells lawyers to research "the program's policies on data retention, data sharing, and self-learning" (Florida Bar). D.C.'s Opinion 388, from April 2024, asks two questions: will what you type "be visible to the GAI provider or other strangers to the attorney-client relationship," and will it "affect answers that later users of the GAI will get" (D.C. Bar). Texas's Opinion 705, from February 2025, calls client information "revealed in responses to future inquiries by third parties" "obviously unacceptable" (Texas Center for Legal Ethics).

Do ChatGPT Enterprise or Claude's commercial terms change the answer?

Yes, on the points the bars care about. Consumer ChatGPT trains on chats unless you opt out and keeps deleted ones for up to 30 days, longer when the law requires it, as we cover in does ChatGPT train on your data and the ChatGPT court order. OpenAI deserves credit for spelling out how business plans differ: it doesn't train on ChatGPT Business or Enterprise data by default, and Enterprise admins set retention (OpenAI, updated January 8, 2026). Its government request policy, effective January 1, 2026, promises the user notice before disclosure "unless prohibited by law" (OpenAI). Anthropic's Commercial Terms likewise say "Anthropic may not train models on Customer Content from Services" (Anthropic). In February 2026, a federal judge held that a defendant's own chats on a consumer Claude account weren't privileged, a case we cover in can AI chats be subpoenaed.

Can a client consent to consumer ChatGPT?

Only with far more than a signature. Under Opinion 512, the client has to hear "the ways in which others might use the information against the client's interests" and "the extent of the risk that later users or beneficiaries of the GAI tool will have access" to it (page 7). For a consumer chatbot, that means explaining training, retention, staff review, and a business-transfer clause that lets personal data be "transferred to a successor or affiliate" in a deal or a bankruptcy (OpenAI, updated September 10, 2026).

D.C. Opinion 388 quotes that same clause, then says clients "are unlikely to give informed consent" and "typically should not be asked to consent" to "wide ranging disclosures that could waive attorney/client privilege." Stripping names doesn't get around the question. D.C. "urge[s] caution" about anonymized prompts, since a public docket or the firm's own website makes the dots easy to connect.

What should a lawyer check in an AI tool before client work?

Opinion 512 borrows its checklist from the earlier cloud computing opinions: confirm the vendor's duty of confidentiality is enforceable, find out what it keeps, and know whether it will tell you about a breach or a demand for your data (page 11). California's 2026 guidance adds that "reasonable efforts require more than reliance on generalized marketing assurances" (State Bar of California, page 5). Here is each requirement mapped to what to check, with consumer ChatGPT's answer as of October 23, 2026.

What does a firm owe when its staff use ChatGPT?

Policies and training. Comment 3 to Rule 5.3 puts "using an Internet-based service to store client information" among the outside services a lawyer must make "reasonable efforts" to keep in line with the lawyer's duties (ABA). Opinion 512 says managerial lawyers "must establish clear policies regarding the law firm's permissible use of GAI," with training on "secure data handling, privacy, and confidentiality" (page 10). D.C. adds that if a tool's privacy policy won't keep client information from outsiders, Rules 5.1 and 5.3 "likely prohibit" letting staff use it on the matter. OpenAI's rule that rating a reply can send the whole chat to training, even after an opt-out, belongs in that training.

How Morse keeps client matters off an AI company's servers

Every row of that table asks about one thing, a copy of your client's information on a vendor's servers under terms the vendor wrote. OpenAI told its zero retention API customers that the preservation order didn't reach their data "because it is not stored" (OpenAI). Morse is built so your chats aren't kept on its servers, and you choose who can read each chat:

• E2EE, for anything that touches a client's matter. Your browser checks the secure enclave's hardware attestation, then encrypts the message so only that enclave can read it, and Morse's servers relay ciphertext they can't read. The enclaves are run by third-party providers, not by Morse, and the guarantee rests on the enclave hardware and attestation working as designed (how it works).
• Private, for research and drafting with no client facts in it. The default. Open-weight models like DeepSeek, Kimi, and GLM, plus Grok, run on providers that have zero data retention agreements with Morse, so the hosts have agreed to keep nothing (what that means).
• Anonymous, for brainstorming with GPT, Claude, or Gemini. Morse's proxy removes your account, email, and IP address. The provider sees the text without knowing it's you and may keep it for a limited time to check for abuse, under its own policy, so use it only for the prompts Opinion 512 says need no consent.

In every mode, Morse doesn't train AI on your content, sell your data, or show ads, so there's no switch and no rating that turns training back on. Its servers don't keep messages or log message content. Your history is stored in your browser, encrypted, and unlocking it takes two keys, one that never leaves your device and one Morse's server gives you at sign-in. Because we don't store chats, we usually have very little to give authorities (privacy policy). Clearing the browser deletes your chats for good, and Morse can't recover them, so copy anything that belongs in the client file into it. Judge all of this against your own rules, reading our privacy policy and terms as Opinion 512 asks you to read any tool's.

Brainstorming with no client information, which Opinion 512 says needs no consent, fits any mode. Client facts, even with names removed, belong in E2EE, and so do strategy and your impressions of the case, which D.C. counts as work product, "among the most protected categories of confidential information."

Try a chat from the Morse home page without an account, or start free with no card needed. End-to-end encrypted chats come with Pro, at $19 a month.