One of the more consequential recent decisions involving AI and legal proceedings came out of a federal case that drew significant attention from legal observers. In United States v. Heppner, the defendant had used an AI chatbot while preparing for litigation. The conversations were later sought by opposing counsel, and the court had to decide whether those communications were protected under attorney-client privilege or the work-product doctrine.
The court’s ruling was clear: AI chatbot conversations do not carry the same legal protections as communications between a client and their attorney. The defendant’s use of AI to work through legal strategy was not shielded from disclosure the way an equivalent conversation with a licensed attorney would have been.
This ruling did not emerge in a vacuum. Courts across the country are grappling with where AI-generated content fits within existing evidentiary and privilege frameworks. Heppner is an early signal that users should not assume their AI conversations are private or protected simply because they “feel” that way. The practical implication is direct: facts you share with an AI chatbot while thinking through your legal situation may be accessible to prosecutors or opposing counsel in ways that genuinely surprise you.
Christopher (Chris) Joseph represents individuals and businesses during the investigation and prosecution of criminal charges in federal and state courts and in complex civil litigation. Chris leads the firm’s criminal defense, asset forfeiture, and judicial discipline practice groups. As JHC’s Managing Member, Chris embodies the firm’s motto: Ready for Anything.
Why This Matters Even If You Don’t Live in New York
Heppner was decided in the Southern District of New York, but federal decisions carry weight beyond the jurisdictions where they originate. While courts in Kansas and Missouri are not bound by every ruling from another federal district, decisions like Heppner reflect a broader legal trend: courts are actively refusing to extend privilege protections to AI interactions used to develop legal strategy. That is a clear signal, regardless of geography.
Kansas and Missouri defendants who use AI chatbots to think through their cases or draft communications are operating under the same fundamental exposure. There is no jurisdiction-specific exception that makes those conversations private. State courts handling criminal matters have broad discretion to consider digital evidence, and AI chat logs are a form of digital evidence.
ChatGPT Is Not Your Lawyer, So Attorney-Client Privilege Does Not Exist
This point seems obvious when stated plainly, but its implications may not be if you do not have a legal background.
Attorney-client privilege is not simply a formal label; it is a legal protection built on a foundation of enforceable duties.
Lawyers owe clients a duty of confidentiality that survives the end of the representation. They are licensed, regulated, and subject to discipline by state bar authorities – discipline that could potentially include disbarment if they violate that duty. Those ethical obligations are codified, monitored, and enforced in every state.
AI systems have none of those characteristics. A chatbot is not licensed. It has no duty of confidentiality. It is not regulated by a bar association, and it cannot be sanctioned for disclosing your information. The conversations you have with an attorney are protected because the law has deliberately chosen to protect them, recognizing that people cannot seek candid legal counsel without assurance of privacy. That policy choice does not extend to software.
What People Are Actually Telling AI
The real risk is not that people are asking abstract legal questions. It is that they are sharing facts: detailed, specific, personally incriminating facts. A review of common AI prompts from people facing legal trouble reveals a consistent pattern:
- “I got arrested last night. Here’s exactly what happened. I told the officer that I was…”
- “Can I delete these text messages before my court date?”
- “How much trouble am I in if I was caught with both the money and the phone?”
- “The witness is someone I know. Is there anything I can do about that?”
- “I used drugs the night before and I wasn’t sure if that would come up in the blood test.”
Each of these prompts contains admissions and context that a prosecutor could use. Information folks would NEVER post on social media, they will enter blithely into an LLM. And that includes things that can land them in hot water later.
Common Misconceptions About Privacy and AI
Many people approach AI chatbots with assumptions about privacy that do not hold up legally. Let’s take a look at some of those assumptions:
“What If My Account Is Private?”
Account privacy settings control what other users can see. They do not restrict what the platform itself can access, store, or be compelled to produce. If a court issues a subpoena or law enforcement obtains a valid legal order, account privacy settings are irrelevant.
“What If I Delete The Conversation?”
Deletion from a user interface does not guarantee that data has been removed from servers. Most AI platforms retain conversation data for safety monitoring, product improvement, and legal compliance. Attempting to delete evidence ahead of a legal proceeding can also carry its own legal consequences, which is a separate and serious problem.
“What If I Pay For ChatGPT/Claude/CoPilot?”
A subscription is a commercial arrangement, not a privilege agreement. Paying for access to an AI tool does not create a confidential relationship in any legal sense. Courts have not treated paid AI subscriptions as analogous to attorney-client or therapist-patient privilege.
Criminal Cases Create Special Risks
Criminal proceedings carry stakes that civil matters often do not. The exposure created by AI chatbot use is more acute here for several reasons.
- Statements can be used as evidence. Anything a defendant writes that reflects their state of mind, knowledge, or actions is potentially admissible. AI conversations are written records.
- Prosecutors seek digital evidence routinely. Law enforcement regularly requests electronic records as part of investigations. AI chat logs are subject to the same requests.
- Defense strategy must remain confidential. Sharing your defense approach with an AI chatbot creates a record that opposing counsel might access.
- The stakes of disclosure are higher. In a civil case, an unfavorable disclosure may cost money. In a criminal case, the consequences extend to liberty.
- AI-generated information can be inaccurate. The well-documented problem of AI hallucination means users may receive confidently stated but factually wrong legal information, influencing decisions in ways that cause real harm.
How to Use AI Responsibly When Looking for Answers
There is a meaningful distinction between using AI tools responsibly and using them in ways that create exposure.
Lower-risk uses include looking up legal terminology, learning how court procedures generally work, or reading about constitutional doctrine. These involve no disclosure of personal facts.
Higher-risk uses include sharing facts about what happened, discussing how to defend against a specific charge given your specific circumstances, describing physical or digital evidence that exists, or sharing anything you would only tell your defense attorney.
That last standard is a reliable test. If you would only share something with a lawyer under privilege, it does not belong in an AI chat window.
Why You Should Call a Lawyer Instead of Using a Chatbot
An attorney provides something an AI tool cannot: a legally protected relationship. The moment you engage a licensed defense attorney, the communications between you become privileged. You can speak candidly about what happened and what questions you have without creating a discoverable record.
That protection is not incidental. It is the foundation of effective legal representation. The defense attorneys at Joseph, Hollander & Craft handle criminal matters across Kansas and Missouri, including felony charges, federal charges, internal investigations, and matters requiring discretion at every stage. No subscription tier, privacy setting, or deleted conversation log provides the protection that an attorney-client relationship does under the law. If you are facing a criminal matter or believe one may be developing, reach out to Joseph, Hollander & Craft to discuss your situation with an attorney.
Quick Questions
Is there any way to use AI safely if I'm already under investigation?
The safest course once you know you are under investigation is to speak with an attorney before taking any additional steps. Anything you type into a chat window at that stage carries increased risk.
Does it matter which AI platform I use?
All major AI platforms retain data to varying degrees and are subject to legal process in their respective jurisdictions. No AI platform offers legally recognized privilege protection comparable to the attorney-client relationship.
What if someone else used AI to research my case without my knowledge?
This raises its own set of concerns depending on what was shared and in what context. An attorney can assess whether that information is legally significant and how it affects your matter.
Could acting on AI-generated legal information hurt my case even if I never shared personal facts?
Yes. Acting on inaccurate legal information can lead to decisions that harm your defense. Relying on AI for legal strategy without attorney oversight creates real risk, separate from any disclosure issue.
What should I do if I already shared detailed facts with an AI chatbot?
Tell us. Do not attempt to delete the conversations before speaking with counsel, as that action could itself create legal complications. We need to know everything to build a proper defense
