Family lawyers across the country are issuing an urgent warning after a February 2025 federal ruling confirmed what many suspected: conversations with AI chatbots like ChatGPT and Claude are not privileged, and opposing counsel in divorce cases can subpoena every word you typed about your finances, custody strategy, or settlement goals. For the estimated 80,000+ divorce cases filed annually in New York, this ruling fundamentally changes how litigants should handle sensitive case information.
Key Facts
| Detail | Summary |
|---|---|
| What happened | Federal court ruled AI chatbot conversations waive attorney-client privilege |
| Key ruling | United States v. Heppner (February 2025) |
| Who reported it | National Law Review / Ward and Smith, P.A. |
| Who is affected | Anyone sharing divorce case details with ChatGPT, Claude, or similar AI tools |
| Discovery risk | Opposing counsel can subpoena full AI chat histories under standard discovery rules |
| New York statute | N.Y. C.P.L.R. § 3101 governs scope of disclosure in civil proceedings |
AI Chat Histories Are Now Fair Game in Divorce Discovery
The United States v. Heppner decision established that sharing confidential information with a consumer AI chatbot destroys any claim of privilege over that information. This is not a gray area. Attorney-client privilege requires communication between a client and a licensed attorney made in confidence for the purpose of obtaining legal advice. ChatGPT is not an attorney, and OpenAI's servers are not a law office. The same applies to Claude, Gemini, Google AI, and every other consumer AI tool on the market.
As reported by the National Law Review, family law practitioners are now actively warning clients that anything typed into an AI chatbot about their divorce is discoverable under standard civil procedure rules. This includes questions about hidden assets, custody positioning, support calculations, and settlement strategy. Opposing counsel who learns that a spouse used ChatGPT for divorce planning can serve a discovery request demanding the complete chat history, and courts will compel production.
The practical danger is significant. A spouse who types "how can I hide $200,000 in assets from my wife during our divorce" into ChatGPT has just created a discoverable record of intent to conceal marital property. That record exists on third-party servers outside the user's control, and no privilege attaches to protect it.
How New York Discovery Rules Apply to AI Chat Histories
New York's discovery framework is exceptionally broad. Under N.Y. C.P.L.R. § 3101(a), parties in a civil action are entitled to "full disclosure of all matter material and necessary in the prosecution or defense of an action." New York courts have consistently interpreted "material and necessary" liberally, requiring only that the information sought be reasonably related to the issues in the case.
In a contested New York divorce proceeding governed by N.Y. Dom. Rel. Law § 236(B), both parties must provide extensive financial disclosure. Equitable distribution under New York law requires the court to consider 14 statutory factors when dividing marital property. If one spouse used ChatGPT to strategize about asset valuation, property concealment, or support calculations, those conversations are directly relevant to financial disclosure obligations.
New York courts also have broad authority to impose sanctions for discovery noncompliance under N.Y. C.P.L.R. § 3126. A party who fails to disclose AI chat histories after a proper discovery demand risks adverse inference rulings, preclusion of evidence, or even dismissal of claims. In a 2024 survey by the American Academy of Matrimonial Lawyers, 35% of family law attorneys reported encountering AI-generated content in divorce proceedings, a figure that has almost certainly grown since.
The subpoena mechanism is straightforward. Under N.Y. C.P.L.R. § 3120, a party can demand production of electronically stored information, which includes chat logs stored on third-party platforms like OpenAI, Anthropic, or Google. Companies receiving valid subpoenas for user data generally comply. OpenAI's privacy policy explicitly states that user data may be disclosed in response to legal process.
What Divorce Clients Are Getting Wrong About AI Privacy
Many people assume that AI conversations are private because they feel private. You are alone with your phone or laptop, typing questions as if speaking to a confidential advisor. That perception is dangerously incorrect for three reasons.
First, AI companies store conversation data on their servers. OpenAI retains ChatGPT conversations unless users specifically opt out, and even then, data may be retained for up to 30 days for safety monitoring. Second, no privilege of any kind attaches to these communications because no attorney-client relationship exists. Third, the information is held by a third party, which under the third-party doctrine means any expectation of privacy is substantially diminished.
New York family courts are particularly attentive to financial disclosure. Under N.Y. Dom. Rel. Law § 236(B)(4), both spouses must file sworn net worth statements. A ChatGPT conversation revealing undisclosed accounts or strategic asset repositioning directly contradicts sworn financial disclosures and could support claims of fraud or contempt.
Practical Takeaways for New York Divorce Litigants
-
Stop sharing case-specific details with any consumer AI tool immediately. This includes ChatGPT, Claude, Gemini, Copilot, and Perplexity. Do not type financial figures, custody goals, settlement positions, or anything about your spouse into these platforms.
-
Assume that every AI conversation you have already had is potentially discoverable. If you used ChatGPT to research divorce strategy before retaining an attorney, inform your lawyer now so they can assess the exposure and develop a response plan before opposing counsel requests production.
-
Use AI tools only for general legal education, not case strategy. Asking ChatGPT "what is equitable distribution in New York" is materially different from asking "how should I structure my equitable distribution claim given my $1.2 million retirement account." The first is general research. The second is case-specific strategy that reveals your financial position.
-
Communicate strategy exclusively through your attorney. Under N.Y. C.P.L.R. § 4503, communications between attorney and client made in confidence for the purpose of obtaining legal advice are privileged. That protection exists nowhere else.
-
Review your AI platform privacy settings. In ChatGPT, navigate to Settings and disable chat history storage. In Claude, check your data retention preferences. These steps do not retroactively protect prior conversations, but they limit future exposure.
Frequently Asked Questions
Can my spouse subpoena my ChatGPT history in a New York divorce?
Yes. Under N.Y. C.P.L.R. § 3101(a), any material relevant to the divorce is discoverable. AI chat histories stored on OpenAI's servers can be obtained through a third-party subpoena under C.P.L.R. § 3120. No privilege protects these conversations because ChatGPT is not a licensed attorney, and no attorney-client relationship exists.
Does deleting my ChatGPT conversations protect me from discovery?
No. Deleting conversations from your visible history does not eliminate them from OpenAI's servers, where data may be retained for up to 30 days after deletion. More critically, intentionally destroying evidence after litigation begins, or is reasonably anticipated, constitutes spoliation under New York law and can result in adverse inference instructions or sanctions under C.P.L.R. § 3126.
What if I only asked ChatGPT general questions about New York divorce law?
General legal research questions carry minimal discovery risk. The danger arises when questions include case-specific facts such as financial figures, asset locations, custody preferences, or settlement targets. A question like "what are grounds for divorce in New York" is benign. A question like "can my husband claim my $500,000 inheritance" reveals specific financial information that opposing counsel can use.
Could AI-generated legal advice hurt my case even if it is never subpoenaed?
Absolutely. AI tools frequently produce inaccurate legal information. ChatGPT may cite repealed statutes, fabricate case law, or misstate New York-specific rules. Acting on incorrect AI advice, such as improperly valuing assets or misunderstanding custody standards under N.Y. Dom. Rel. Law § 240, can result in unfavorable court outcomes that are difficult to reverse on appeal.
Is there any way to use AI safely during a divorce?
AI tools are safe for general education about divorce processes, terminology, and publicly available legal information. The critical boundary is never inputting case-specific facts, financial data, or strategic considerations. Treat every AI chatbot as a public forum where opposing counsel is reading over your shoulder. For any case-specific question, consult a licensed family law attorney whose communications are protected by N.Y. C.P.L.R. § 4503 privilege.
If you are navigating a divorce in New York and have questions about discovery obligations or protecting sensitive information, connect with a qualified family law attorney in your county through our New York directory.
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.