When a legal question pops into your head at 11 p.m. or you’re rehashing a detailed conversation you had with your attorney, it's tempting to type the whole story into your favorite AI tool and see what comes back.
AI chatbots can be a great resource. In certain circumstances. For certain things.
But, in these contexts- before you do, it’s worth pausing. One of the most valuable protections you have as a client, and one that’s easy to give away unintentionally, is attorney-client privilege.
Why privilege is worth protecting
Attorney-client privilege is one of the quiet workhorses of the legal system. It lets you tell your lawyer the whole story; including the awkward parts, and the part you may not want the public to be privy to, knowing the details stays between the two of you. That candor is what makes good advice possible. A lawyer can only develop a strong strategy after hearing the real facts- all of them.
For business owners, creators, and inventors, privilege does a lot of everyday work. It lets you talk openly about whether a new brand name might step on someone’s toes, whether a contract is as friendly as it looks, or how to handle a dispute before it turns into a lawsuit. It gives you room to think out loud, ask “what if” questions, and plan strategy without worrying that your notes will become the other side’s Exhibit A.
The protection is strong, and it comes with a few simple conditions: a real lawyer, a legal purpose, and confidentiality. The most common way to lose it is sharing the conversation with someone outside that circle. That’s where AI may enter the picture.
What one court said
In February 2026, a federal judge in New York took up this question in United States v. Heppner. A criminal defendant had used a consumer AI chatbot to work through his legal situation, sent the results to his lawyers, and then claimed the documents were privileged.
The Judge said no, and his reasoning was simple: there is no attorney-client or work product protection for documents a client created using the generative AI chatbot and subsequently sent to his lawyer.
A chatbot isn't a licensed attorney, so there's no attorney-client relationship. The platform's consumer terms didn't promise confidentiality, so there was no reasonable expectation of privacy.
Further, the court also accepted the idea that pasting your lawyer's advice into a public AI tool can waive privilege over that advice. That part is worth sitting with for a minute.
Not doom, just nuance
None of this means AI and legal work can't get along. The court noted the result might be different if a lawyer directs a client to use a specific tool as part of the representation. Some commentators also see room for AI to act like the other outside helpers lawyers already rely on, such as accountants or translators.
Business-grade tools with real confidentiality terms may be a different story too. Courts are still working it out, so proceeding on the side of caution is vital.
The other catch: confident, and sometimes wrong
Privilege isn't the only wrinkle. AI tools can hallucinate, which is the polite term for just making things up. They can invent court cases that don't exist, or cite real cases for points those cases never made, in the same confident tone they use for everything else.
This isn't hypothetical. In January 2026, a federal judge in Pennsylvania sanctioned two lawyers in an IP matter over at least eight AI-invented citations. In June 2026, a Mississippi federal judge sanctioned lawyers on both sides of the same case, canceled the trial, and barred one attorney from that court for two years. If it can happen to an attorney, it can happen to you.
Even when the citations are real, the reasoning can be flawed. AI can misread a holding, blend rules from different states, or miss that the law changed last month. An AI answer can sound like a legal opinion without being one.
A few questions before you hit enter:
What does this tool's fine print say about confidentiality and data sharing?
Would I be comfortable if this chat became public or showed up in a lawsuit?
Am I pasting in something confidentially discussed with my attorney, or advice provided by my attorney?
Am I describing an unfiled invention or an unannounced brand?
If I'm relying on an AI answer, has a human checked that the cases exist and actually support what it claims?
Think of a public chatbot as a smart, very chatty acquaintance who occasionally makes things up and won't fess up to it unless you know enough to question the answer. Good for brainstorming and providing general information or lists of questions on a legal topic. Not where you make your confessions.
Let's talk (privately)
Reach out anytime. I promise our conversations stay between us*.
*with certain, very specific exceptions.
References & Additional Reading:
Federal Court Rejects Privilege Assertion for AI-Generated Research, American Bar Association, Litigation Section
AI Chatbots, Privilege, and Pitfalls, Goodwin Procter (March 2026)
US Court: AI Chats Are Not Privileged, The Next Web (April 15, 2026)
Negating Attorney-Client Privilege: Don't Let AI Put a Fox in Your Company's Henhouse, Mondaq (March 30, 2026)
Lawyers Sanctioned Over AI Use on Both Sides in Federal Case, Bloomberg Law (June 9, 2026)
Judge Aycock AI Hallucination Sanctions, Lawyer Monthly (June 2026)
Two More Lawyers Punished for AI Misuse, Including $4,000 Fine, Bloomberg Law (January 27, 2026)
AI and Judicial Sanctions, Temple Law 10Q (August 12, 2026)