Can asking ChatGPT about my estate plan create legal risk?

Legal disclaimer: This article is for general informational and editorial purposes only. It is not legal advice, does not create an attorney–client relationship, and should not be relied on as a substitute for advice from a licensed attorney in your jurisdiction. Privilege, confidentiality, discovery, and estate-planning rules vary by jurisdiction and by the facts of each matter. Nothing here is a prediction of how any court would treat a particular conversation, document, or tool.

Generative AI tools such as ChatGPT can help people organize questions, learn vocabulary, and sketch high-level topic lists before meeting an estate-planning attorney. They can also create real risk when someone pastes wills, trusts, beneficiary details, net-worth schedules, family conflict notes, or strategy into a consumer chatbot and treats the output as confidential legal guidance.

This Haute Lawyer Legal Questions explainer frames confidentiality and privilege at a general level, notes accuracy limits, flags sensitive-data leakage, and clarifies when AI is a preparation aid versus when licensed counsel is required.

Short answer

Yes. Asking a consumer AI tool about your estate plan can create legal and privacy risk, especially if you paste in documents, financial details or your lawyer's advice. A chatbot isn't your lawyer, and chats with it are generally not privileged. In February 2026, a federal judge in Manhattan held that a defendant's exchanges with Anthropic's Claude were protected by neither attorney–client privilege nor the work-product doctrine (United States v. Heppner, S.D.N.Y.). Use AI, if at all, for vocabulary and question lists. Keep drafts, numbers and counsel's emails in your lawyer's secure channels.

Why estate planning is a high-sensitivity topic for AI chats

Estate planning often involves information people would never post publicly:

• Family structure, incapacity concerns, and conflict dynamics
• Asset lists, business interests, and liquidity
• Beneficiary designations and unequal distributions
• Trust design themes and tax-planning questions
• Draft wills, trust instruments, or letters of intent

Putting that material into a third-party consumer AI system can create confidentiality, privilege, accuracy, and discovery-related concerns that a casual "just brainstorming" mindset underestimates.

Confidentiality and privilege: general risk framing

Attorney–client privilege is narrow

In U.S. practice, attorney–client privilege generally protects confidential communications between a client and a licensed attorney made for the purpose of obtaining or providing legal advice—subject to elements and exceptions that vary by jurisdiction. A consumer chatbot:

• Is not a licensed attorney
• Does not form an attorney–client relationship with you
• Typically disclaims that it provides legal advice

So a chat with ChatGPT (or a similar consumer generative AI tool) is generally not privileged merely because the topic is "legal."

Consumer AI and confidentiality expectations

Privilege and confidentiality analyses often turn on whether a communication was intended to be—and in fact was—kept confidential. Consumer AI platforms commonly collect inputs and outputs under their terms and privacy policies, may use data for model improvement (depending on product tier and settings), and may reserve rights to disclose information to third parties in defined circumstances. Users should read the applicable terms for the exact product and account type they use.

Practical takeaway: Do not assume a private-feeling chat window equals a confidential lawyer conference.

A public federal ruling to know (fact-specific—not a national "ChatGPT ban")

In United States v. Heppner, No. 25 Cr. 503 (JSR), ECF No. 27 (S.D.N.Y. Feb. 17, 2026), Judge Jed S. Rakoff of the U.S. District Court for the Southern District of New York ruled from the bench on February 10, 2026, and followed with a written memorandum on February 17, 2026. He held—on the facts of that case—that a defendant's exchanges with a consumer generative AI platform (Claude, operated by Anthropic) were protected by neither the attorney–client privilege nor the work-product doctrine. The court relied on three points: the AI provider is not an attorney, the platform's privacy policy defeated any expectation of confidentiality, and counsel had not directed the searches. A footnote adds that sharing otherwise privileged information with the AI platform could waive privilege as to that information. The court called the question "a question of first impression nationwide."

How to read that carefully:

• It is a real, attributed federal district-court decision—not invented case law.
• It is fact-specific (consumer platform; user-initiated use; not counsel-directed on the facts described).
• It involved a different consumer AI platform than ChatGPT (Claude, not ChatGPT), but the risk themes—third-party disclosure, lack of an attorney relationship, and waiver concern—are the reason estate-planning clients should be cautious with any consumer generative AI chat about sensitive legal facts.
• It is not a holding that every AI use everywhere is unprotected, nor that enterprise tools under counsel's direction would be analyzed identically. Outcomes can differ with different facts, platforms, contracts, and courts.

What your ChatGPT settings do and don't change

OpenAI's Help Center describes the training and retention controls for ChatGPT:

• Turning off "Improve the model for everyone" stops new conversations from being used to train OpenAI models, but saved chats stay in your history.
• Temporary chats are not used to improve OpenAI models, but may be retained for up to 30 days for safety purposes.
• Giving a thumbs-up or thumbs-down on a response can put the entire conversation associated with that feedback into training, even if you opted out.
• By default, OpenAI does not use content from ChatGPT Business, Enterprise, Edu, or ChatGPT for Healthcare workspaces to train its models.

These settings concern training and retention. They don't make a chat privileged.

Accuracy limits: fluent is not correct

Even when privacy is not the main worry, estate planning is a poor fit for unverified AI drafting because models can:

• Invent statutes, cases, forms, deadlines, tax rules, or "standard" clauses that do not apply in your state
• Miss interactions among wills, trusts, beneficiary designations, titling, and probate alternatives
• Overlook capacity, undue-influence, and execution formalities that determine whether documents work when it matters
• Sound confident while being wrong—especially on multi-state or high-net-worth fact patterns

An AI outline of "what people often put in a will" is not a substitute for counsel who can apply your facts to your jurisdiction's requirements.

Sensitive-data leakage: what can go wrong in plain terms

1. Third-party access under platform terms — Inputs may be stored, reviewed, or disclosed under the provider's policies.
2. Account and device risk — Shared laptops, synced browsers, workplace monitoring, or compromised accounts can expose chat history.
3. Downstream human sharing — Screenshots, exports, and "I asked ChatGPT and it said…" emails can spread sensitive details to relatives, advisors, or adversaries.
4. Discovery and disputes — Written materials about estate strategy, family conflict, or asset lists can later become relevant in litigation, audits, or contested-probate settings. Treat AI chats as potentially discoverable or leaky, not as sealed attorney files.
5. Waiver themes — Sharing advice you already received from counsel into a consumer AI tool can raise waiver concerns regarding that underlying privileged information. Ask your own attorney before pasting counsel's emails or memos into any chatbot.

When AI use may be lower-risk vs when to use counsel

Often lower-risk (still not advice)

• Learning public vocabulary ("what is a pour-over will?") from educational prompts that contain no personal identifiers
• Building a question list for your first attorney meeting
• Summarizing public articles or government explainers you already have permission to use
• Organizing a personal checklist of documents to gather—without uploading the documents themselves into a consumer AI tool

Higher-risk / usually avoid for consumer AI

• Pasting draft wills, trusts, or redlines
• Uploading net-worth schedules, cap tables, or tax returns
• Describing specific family disputes, capacity concerns, or unequal-bequest rationales with identifiable detail
• Asking the model to "decide" distribution percentages or trustee choices for your real family
• Feeding the model emails or memos from your attorney
• Relying on AI to execute, notarize, or "finalize" estate documents. AI can't sign or witness a will. In Florida, for example, a will must be signed at the end by the testator in the presence of at least two attesting witnesses, who also sign (Fla. Stat. §732.502). Other states have their own rules.

When to speak with licensed counsel

Consider an estate-planning attorney sooner if you:

• Own a business, have multi-state property, or have a blended family
• Need trusts, tax coordination, or special-needs planning
• Have recently married, divorced, had children, or received a large inheritance
• Are updating documents after a major life or asset change
• Have received a diagnosis or face capacity/elder-care decisions
• Are unsure whether existing beneficiary designations match your intended plan

Early counsel is often about document hygiene and process, not only "writing a will this afternoon."

Ask how your estate lawyer uses AI

The question runs both ways. ABA Formal Opinion 512 (July 29, 2024) says a lawyer needs the client's informed consent before inputting information relating to the representation into a self-learning generative AI tool, and that boilerplate language in an engagement letter is not sufficient. State bars adopt their own rules, so ask your estate lawyer whether, when and how they use AI tools with your information.

Featured estate planning attorneys

Michael Kosnitzky · Robert W. Haley

• Michael Kosnitzky: partner and co-leader of Pillsbury Winthrop Shaw Pittman LLP's Private Wealth practice. His profile describes an approach integrating tax planning, estate planning, asset protection and family governance. Admitted in Florida, New York, the District of Columbia and Colorado, with offices in Miami and New York.
• Robert W. Haley: Certified Elder Law Attorney (CELA®) and Managing Attorney of The Estate & Elder Law Center of Central/Southside Virginia, PLLC. Licensed in Virginia, with offices in Danville and Lynchburg. His profile lists estate planning, Medicaid planning and asset protection.

Haute Lawyer membership is a paid editorial program; inclusion is not a ranking or recommendation. See also Estate Planning/Probate attorneys.

Related Haute Lawyer pages

• How to Choose an Estate Planning Attorney for a High-Net-Worth Family
• Estate Planning for High-Net-Worth Families
• How Often Should You Update an Estate Plan?
• What Is a Beneficiary Designation?
• Questions to Ask an Estate Planning Attorney

Key takeaways

• Asking consumer AI about an estate plan can create confidentiality, privilege, accuracy, and leakage risks—especially when personal facts or documents are pasted into the chat.
• Consumer chatbots are not attorneys; chats with them are generally not attorney–client privileged communications.
• United States v. Heppner (S.D.N.Y. Feb. 17, 2026) is a fact-specific public ruling rejecting privilege/work-product claims for certain consumer AI materials.
• Use AI, if at all, for vocabulary and question lists; keep drafts, numbers, and counsel communications inside secure attorney channels.
• Haute Lawyer Legal Questions is editorial Q&A—not legal advice, not rankings, and not a substitute for licensed counsel.