People confide in AI chatbots the way they used to confide in a diary. They ask ChatGPT how to respond to a spouse during a divorce, or type out the details of a custody dispute at midnight when their attorney's office is closed. Typing feels private, almost like thinking out loud.
Federal courts have started saying otherwise. Earlier this year, a judge in New York ruled that a defendant's conversations with an AI platform were not protected by attorney-client privilege, even though he later shared the AI's output with his own lawyers. The ruling landed as one of the first of its kind in the country, and judges in other states are now weighing similar questions.
The early answers all point in the same direction: what gets typed into an AI tool about a legal matter does not automatically stay between the person and the screen, and treating it that way can carry real consequences for a case.
The Ruling That Put This Risk on the Record
In United States v. Heppner, a federal court in the Southern District of New York became the first in the country to address whether a person's conversations with an AI chatbot could be shielded by attorney-client privilege. The answer was no.
Bradley Heppner had been indicted on federal fraud charges. According to court filings, after learning he was likely to be charged, Heppner used the AI assistant Claude on his own, without instruction from his attorneys, to organize his thinking and draft reports outlining possible defense arguments. He later shared what Claude produced with his legal team.
When prosecutors seized Heppner's devices, they found those exchanges and asked the court to rule that the material was fair game. The court agreed, for three reasons:
- Claude is not a licensed attorney, so no attorney-client relationship could have formed in the first place.
- The exchanges were not confidential, since Heppner had communicated with a commercial AI platform whose privacy policy permits data collection and disclosure to outside parties, including government authorities.
- Heppner had not used Claude at his attorneys' direction, so the court found he could not have intended to obtain legal advice from the platform itself, only from the humans he worked with afterward.
The court also rejected a separate argument that the material qualified as protected work product prepared in anticipation of litigation, since Heppner's own attorneys had neither directed nor prepared it. The outcome could differ if an attorney specifically directs a client to use an AI tool in a particular way. Absent that direction, the privilege did not apply.
Why AI Chats Feel Private When They Are Not
Part of what makes the Heppner ruling surprising to most people is how the interface feels. Typing into a chat window feels like private thinking. In reality, every exchange becomes a timestamped, searchable record sitting on a company's servers, governed by that company's own privacy policy rather than anything resembling attorney-client confidentiality.
Under the privacy policies used by many consumer AI services, providers may retain what users submit, use some conversations to maintain or improve their services depending on the user’s settings, and disclose information when required by law or valid legal process, such as a subpoena.
AI Is Also Getting Attorneys in Trouble
Clients are not the only ones facing consequences. The trend traces back to a 2023 case out of New York, where an attorney became a cautionary tale nationwide after citing court decisions that ChatGPT had invented for a filing. Since then, courts across the country have been sanctioning lawyers who filed briefs containing case citations and quotes that AI tools invented outright. A few recent examples:
- Two Orange County attorneys were fined and suspended from practicing before the court for six months after submitting case briefs citing opinions that did not exist.
- An attorney and his firm were ordered to pay over $10,000 in sanctions after a civil brief contained fabricated case citations the attorney admitted he never verified.
- A federal judge suspended an attorney from practicing in the district for six months over fabricated citations in a court filing.
- A federal judge disqualified attorneys from a case entirely after filings on both sides turned out to contain fabricated citations.
- One federal appeals court handed down a $30,000 sanction this year, believed to be the largest federal appellate penalty tied to fabricated AI citations to date.
The lesson for anyone hiring a lawyer is straightforward: ask how a firm verifies its own research before it becomes part of your case.
How to Keep Your Case Protected
If a legal matter is already underway, or about to begin, a few habits keep the protections designed to work in your favor:
- Do not type details about a legal matter into ChatGPT, Claude, Gemini, or any other public AI tool without talking to your attorney first.
- Do not ask AI to help draft messages to a spouse, co-parent, or opposing party in an active dispute. AI-drafted messages like these can become evidence of intent or state of mind later in the case.
- Do not summarize what an attorney told you in confidence, then paste that summary into an AI tool to think it through. Doing so cost Heppner his privilege claim.
- Bring questions straight to an attorney. A phone call or message to counsel carries protections an AI conversation does not.
- Ask, if AI comes up at all in a case, whether the firm handling the matter has a policy for how it verifies AI-assisted research before anything reaches a filing.
A Firm Built Around Direct Communication with Clients
People going through a divorce, a custody dispute, or an estate matter still need somewhere to bring their questions the moment those questions come up. Vaclavek Hartman Vaclavek was built around exactly that need, serving McHenry and Kane County families since 2017 with family law, estate planning, and real estate representation.
What that looks like in practice:
- Direct access to the attorney handling the case. Clients working with Vaclavek Hartman Vaclavek reach the person actually representing them rather than routing questions through layers of staff, so a pressing question gets answered by someone who already knows the case.
- Free consultations, offered virtually or in person. Anyone weighing whether to move forward with a divorce, a will, or a real estate transaction can sit down with an attorney and ask direct questions before committing to anything, at no cost and on a schedule that fits around work and family obligations.
- Personalized strategy instead of a one-size-fits-all approach. High-conflict divorces, high-net-worth divorces, and contested estate matters each carry their own pressure points, and Vaclavek Hartman Vaclavek builds a strategy around the specific facts of each case rather than applying the same template to every file.
- Support that goes beyond paperwork. Family law and estate matters are often difficult to navigate alone, and Vaclavek Hartman Vaclavek checks in proactively and explains next steps in plain language, so clients are never left guessing what happens next.
Whether the matter is a divorce, a custody dispute, an estate plan, or a real estate transaction, the safest place to think out loud about it is with an attorney, not an AI tool. Call (815) 507-8887 or reach out to Vaclavek Hartman Vaclavek online to schedule a free consultation.