If you use a commercial AI chatbot – such as ChatGPT, Claude, or Gemini – to discuss, record, summarize, or strategize about legal matters, a federal court has now confirmed that those conversations are not protected by attorney-client privilege.
Even sharing the AI output with your attorney after the fact does not restore the privilege.
Why This Ruling Matters for Texas Business Owners and Healthcare Practitioners
AI tools are transforming how business owners and healthcare practitioners manage their day-to-day operations – and that includes legal matters. You may already use tools like ChatGPT, Claude, or Gemini to draft contracts, summarize meetings with your attorney, outline compliance strategies, or brainstorm next steps for a dispute.
If so, you need to pay attention to a recent federal court ruling that could leave your most sensitive legal conversations completely unprotected.
On February 10, 2026, Judge Jed Rakoff of the U.S. District Court for the Southern District of New York ruled in United States v. Heppner that documents a defendant created using a consumer AI platform were not protected by attorney-client privilege or the work product doctrine.
The court’s reasoning has implications that reach well beyond criminal cases – and directly into the conference rooms, clinics, and offices of Texas business owners and healthcare providers.
What Are the Three Requirements for Attorney-Client Privilege?
Attorney-client privilege has three essential elements. A communication must be:
- Between a client and their attorney;
- Intended to be and actually kept confidential; and
- Made for the purpose of obtaining or providing legal advice.
Under Texas Evidence Rule 503, a communication is confidential only if it is “not intended to be disclosed to third persons” other than those helping to deliver legal services. Federal courts apply a similar standard.
When you input information into a commercial AI tool whose privacy policy permits data collection, model training, and third-party disclosure, you have disclosed that information to a third party and may have forfeited attorney-client privilege. Once lost, the privilege does not come back.
What Happened in United States v. Heppner?
Bradley Heppner, the CEO of a financial services company, faced federal fraud charges including securities fraud, wire fraud, and falsification of records. After receiving a grand jury subpoena and retaining defense counsel, Heppner used the publicly available AI tool Claude to research and analyze his legal situation. He created approximately 31 documents through his interactions with the platform.
When federal agents seized Heppner’s devices during a search, they discovered these AI-generated documents. Heppner’s defense team argued the documents were privileged – that some of the information Heppner typed into the AI tool came from conversations with his attorneys, and that he later shared the AI outputs with his legal team.
The court rejected these arguments on three grounds.
1. AI Documents Are Not Communications with Your Attorney
Attorney-client privilege protects confidential communication between a client and an attorney made for the purpose of obtaining legal advice. The court found that communications between Heppner and an AI platform simply do not qualify.
An AI tool cannot form an attorney-client relationship with you. It is not licensed to practice law, does not owe you fiduciary duties, and cannot be disciplined for misconduct. Discussing legal issues with an AI chatbot is no different – legally speaking – from discussing them with a friend or colleague. Neither conversation is privileged.
2. Your AI Conversations Are Not Confidential
Confidentiality is the cornerstone of attorney-client privilege. The court examined the AI platform’s privacy policy and found that the provider:
- Collects both user inputs (your prompts) and AI outputs (the responses)
- Uses that data to train and improve its AI models
- Reserves the right to disclose data to third parties, including government authorities – even without a subpoena
Heppner could have had no “reasonable expectation of confidentiality in his communications” with Claude. Uploading confidential notes to an AI tool is like sharing
When you type sensitive legal information into a commercial AI tool that uses your data for training, you are effectively sharing that information with a third party. Courts have long held that once information is in the hands of a third party, attorney-client privilege is lost.
In the Heppner case, the judge reasoned, uploading confidential notes to an AI tool like Claude is equivalent to discussing your legal strategy in a crowded room.
3. Sharing AI Output with Your Lawyer Does Not Restore Privilege
Heppner argued that because he eventually shared the AI-generated documents with his attorneys, the privilege should apply. The court disagreed.
Under well-established law, sending non-privileged documents to your attorney does not make them privileged after the fact. The court was clear: non-privileged communications are not “alchemically changed into privileged ones upon being shared with counsel.”
Warner v. Gilbarco – When AI Use Did Not Waive Privilege
The same week as the Heppner ruling, a federal court in Michigan reached a different outcome in Warner v. Gilbarco. In that case, the court held that internal materials created using generative AI tools are protected by the work-product doctrine (instead of attorney-client privilege).
Work-Product Doctrine Vs. Attorney-Client Privilege
The work product doctrine protects documents and materials prepared by attorneys or their representatives in anticipation of litigation from disclosure during discovery. Unlike attorney-client privilege, which protects confidential communications between lawyers and clients, the work product doctrine protects all materials prepared for litigation, not just communications.
In the Warner ruling, court explained that “while … a voluntary disclosure to a third person will generally suffice to show waiver of the attorney-client privilege, it should not suffice in itself for waiver of the work product privilege.” Third-party disclosure only waives work-product protection if it significantly increases the likelihood that an adversary (not just any third-party, but specifically an adversary) will see the material.
Another critical difference? In Warner, the AI tools were used by the pro se plaintiff (i.e., she acted as her own counsel) to prepare her case. In Heppner, the defendant acted on his own, without his attorney’s direction.
The takeaway: Work-product protection only applies to materials prepared "at the behest of counsel". Since Heppner used an AI tool on his own initiative, the work-product protection never attached.
How Does This Apply to Your Business or Practice?
You do not need to be a defendant in a federal criminal case for this ruling to affect you. Consider these real-world scenarios.
For Texas Business Owners
- Partnership dispute: You suspect your business partner is diverting company funds. You open ChatGPT and type detailed questions about your legal options, including facts your attorney shared with you. If litigation follows, the opposing party’s attorneys may argue – with support from Heppner – that those AI conversations are discoverable.
- Contract negotiation: You use an AI tool to transcribe a confidential consultation with your attorney, allowing it to summarize the conversation and ask follow-up questions. The transcription and summary now exist on a third-party server, potentially accessible to the AI provider and, by extension, to others.
- Employment matter: You ask an AI chatbot to help you draft a response to a former employee’s demand letter, inputting details of your attorney’s advice. That privileged strategy may no longer be protected.
For Healthcare Practitioners
- Board investigation: You receive notice of a Texas Medical Board investigation and use a consumer AI tool to research your legal exposure, inputting details about the complaint and your attorney’s guidance. Under the Heppner framework, those interactions could be fair game for the investigating body.
- Regulatory compliance: You use an AI tool to analyze whether your practice’s billing arrangements comply with the federal Anti-Kickback Statute, providing detailed information about your compensation structures. If a government investigation follows, that information could be discoverable.
- Practice acquisition: You’re negotiating the purchase of another practice and use AI to organize confidential due diligence findings your attorney prepared. Those summaries may lose their privileged status the moment they enter the AI platform.
How to Protect Attorney-Client Privilege When Using AI
When you type confidential information into an AI platform, are you disclosing it to a third party – or simply using a tool to process your own work? Courts are beginning to answer that question, and the consequences for your legal protections are significant.
Until courts provide more clarity, the safest approach is straightforward: do not use consumer AI tools to discuss, record, summarize, or research the specific facts of your legal matter. Do not input details your attorney shared with you. Do not ask an AI chatbot to suggest next steps for a pending dispute, investigation, or transaction.
If you want to use AI to assist with legal work, talk to your attorney first. The right safeguards – including enterprise-grade tools, counsel direction, and contractual confidentiality protections – can make a meaningful difference in whether your communications remain protected.
AI tools offer real value to businesses and healthcare practices. The goal is not to avoid them. It is to use them wisely, with an understanding of where the legal boundaries are.
If you have questions about how AI use may affect your legal protections – or if you need help developing AI usage policies for your business or practice – the attorneys at Hendershot Cowart P.C. can help. Call (713) 783-3110 to schedule a consultation.
Frequently Asked Questions
Does this ruling apply in Texas?
Heppner is a federal court decision from the Southern District of New York. While it is not directly binding on Texas courts, the legal principles it applies – including the third-party disclosure waiver rule – are well-established across federal and Texas state courts. Texas Evidence Rule 503 imposes similar confidentiality requirements for privilege to attach.
What if the AI tool says it does not store my data?
Read the privacy policy carefully. Many consumer AI tools still collect, process, and retain data even when they claim not to use it for training. Look for explicit contractual commitments about data handling, not just marketing language. When in doubt, treat the platform as a third party.
Can I use AI for general legal research without risking privilege?
Using AI to research general legal topics – such as understanding how non-compete agreements work in Texas – is unlikely to raise privilege concerns because you are not sharing privileged information. The risk arises when you input facts, strategies, or communications related to your specific legal matter.
Can my attorney use AI tools?
If your attorney uses an enterprise-grade AI platform that does not train on your data and provides contractual confidentiality protections as part of their legal representation, privilege is likely protected based on the outcome of the Heppner case. The key factors are whether the tool maintains confidentiality and whether its use is directed by counsel. Discuss this directly with your attorney.