A federal court held in U.S. v. Heppner on February 17, 2026 that materials created with public AI tools like ChatGPT are not protected by attorney-client privilege or work-product doctrine. For New York divorce and custody litigants, this means the AI conversations you had about hidden assets, finances, or settlement strategy can be compelled during discovery — and paid AI tiers offer no legal protection.
Family law firms across the country began issuing warnings in mid-June 2026 after the Ward and Smith analysis flagged the ruling's discovery implications. The practical takeaway is blunt: anything you type into ChatGPT about your divorce is potentially discoverable evidence your spouse's attorney can demand to see.
| Key Fact | Detail |
|---|---|
| What happened | A federal court ruled AI-tool chats (e.g., ChatGPT) are not covered by attorney-client privilege or work-product doctrine |
| When | Decided February 17, 2026; family-law warnings escalated mid-June 2026 |
| Where | U.S. federal court ruling; principle applies to state divorce discovery nationwide, including New York |
| Who's affected | Anyone using public AI tools to discuss divorce finances, custody, hidden assets, or settlement strategy |
| Key rule | Attorney-client privilege requires confidential communication with a lawyer; AI vendors are third parties that break confidentiality |
| Practical impact | Spouses can be compelled to produce AI chat logs during discovery; paid AI tiers provide no privilege protection |
Why this matters legally
U.S. v. Heppner confirms that attorney-client privilege does not attach to conversations with an AI chatbot. Privilege protects confidential communications between a client and their lawyer for the purpose of obtaining legal advice — and a chatbot operated by a commercial vendor is neither your lawyer nor a confidential channel. When you share information with a third party like OpenAI, you waive any expectation of confidentiality, which is the foundation privilege rests on.
The work-product doctrine fares no better. Work product protects materials prepared by or for an attorney in anticipation of litigation. A spouse drafting questions to ChatGPT is not an attorney preparing a case, and the AI vendor is not acting under attorney direction. The ruling treats these chats as ordinary communications with a third-party service, fully exposed to discovery.
This is a definitive shift, not a gray area. If you used ChatGPT to ask how to hide a brokerage account, calculate what you could lowball in a settlement, or strategize custody positioning, those logs are now squarely within the scope of what an opposing party can request. Many AI platforms also retain conversation history and respond to lawful subpoenas, meaning deletion after a dispute begins can itself create spoliation problems.
How New York law handles this
New York divorce discovery is built for broad disclosure, which is exactly why the Heppner principle bites hard here. Under N.Y. C.P.L.R. § 3101, parties are entitled to full disclosure of all matter material and necessary to the case. That standard is interpreted liberally — if your AI chats discuss assets, income, or parenting, they are presumptively discoverable unless a recognized privilege applies, and Heppner says privilege does not apply.
Financial transparency is mandatory in New York matrimonial actions. N.Y. Dom. Rel. Law § 236 governs equitable distribution and requires both spouses to file a sworn Statement of Net Worth disclosing all assets, debts, income, and expenses. Courts can impose sanctions for concealment. An AI chat log showing a spouse asking how to move money before filing is the kind of admission that can devastate an equitable-distribution position and undermine credibility on every other issue.
Custody adds another layer. New York courts decide parenting under the best-interests standard codified in N.Y. Dom. Rel. Law § 240. A discoverable AI conversation where a parent describes coaching a child, manufacturing allegations, or gaming the custody process can be used to question that parent's judgment. Because New York's disclosure rules reach broadly, there is no reliable technical or legal shield once these chats exist.
Practical takeaways
-
Stop using public AI tools to discuss your divorce. Do not type questions about hidden assets, settlement floors, custody strategy, or your spouse into ChatGPT, Gemini, Claude, or any consumer chatbot. Treat every prompt as a potential exhibit.
-
Do not delete existing chats once a dispute is foreseeable. Deleting evidence after litigation is anticipated can trigger spoliation sanctions under New York's discovery framework. Tell your attorney what exists and let them advise on preservation.
-
Move sensitive questions to your actual attorney. Communications with a licensed lawyer for legal advice are protected by attorney-client privilege. That protection is what you are paying for — a chatbot cannot replicate it, regardless of the subscription tier.
-
Assume paid tiers offer no privilege. Heppner draws no distinction between free and paid AI services. Paying OpenAI does not create a confidential legal relationship; the vendor remains a discoverable third party.
-
Audit your digital footprint early. When you retain counsel, disclose AI tool usage, cloud documents, and messaging apps so your attorney can assess exposure under N.Y. C.P.L.R. § 3101 before opposing counsel does.
If you are navigating a New York divorce and worried about what your past research or communications might reveal, the safest step is to bring those concerns to a qualified family law attorney who can evaluate them under privilege — not to a chatbot. An experienced matrimonial lawyer can help you understand your discovery obligations and protect your position.
This article discusses recent news and provides general legal commentary. It does not constitute legal advice. Every case is unique. Consult a qualified family law attorney for advice specific to your situation.