Should I Use AI to Help Me Communicate With My Criminal Defense Attorney in Pennsylvania?

If you’re facing criminal charges or are under investigation, you may have a head full of questions and very little idea how to organize them. It's natural to open ChatGPT, Claude, or another AI chatbot and start typing. Maybe you want help writing a timeline of what happened, summarizing the police report, or drafting an email to your lawyer so you don't forget anything important.

A federal judge in New York has held that those conversations with AI chat bots are not private or protected from disclosure. In United States v. Heppner, the court held that a defendant's conversations with a public AI chatbot about his own case were not protected by the attorney-client privilege or the work-product doctrine, and the government was allowed to use that information at trial.

Our firm’s advice on this issue is simple - DO NOT use AI to prepare to talk to your lawyer, or to assist with your defense, unless you are specifically instructed to use AI by your attorney.

What Happened in Heppner

Heppner, a former corporate CEO and board chairman, received grand jury subpoenas and learned he was a target of a federal investigation. He hired a lawyer and was later indicted on securities and wire fraud charges.

Before the indictment, and without being told to by his attorney, Heppner used a publicly available consumer AI platform (reported to be Anthropic's Claude) to organize his thoughts and information he believed was relevant to his defense. Some of what he typed reflected things his lawyers had told him. He used the tool to generate written materials meant to help him prepare for future conversations with his attorneys.

When Heppner was arrested, the FBI searched his home and seized his devices. Those devices held 31 documents generated through his AI sessions. His lawyers claimed the documents were privileged. The government asked the court to rule that they weren't. Judge Jed Rakoff sided with the government, ruling from the bench on February 10, 2026 and issuing a written opinion a week later. The court described the question as one of first impression nationwide.

Why the Court Said the AI Conversations Weren't Privileged

The attorney-client privilege protects confidential communications between a client and a lawyer made for the purpose of getting legal advice. The court found that Heppner's AI conversations failed that test for three independent reasons:

The chatbot is not a lawyer. The privilege rests on a relationship with a licensed professional. The court compared typing into an AI tool to using ordinary cloud-based software, which is not automatically privileged.

• There was no reasonable expectation of confidentiality. The platform's privacy policy said it collected users' inputs and outputs, used them to train its model, and could disclose them to third parties, including government authorities. Under those terms, the court concluded Heppner could not reasonably expect his conversations to stay private.

• He wasn't seeking legal advice from the chatbot. The tool's own terms said it is not a lawyer, and when asked, it said it could not give legal advice.

The defense argued that Heppner meant to share the documents with his lawyers afterward. The court rejected that too. A communication that wasn't privileged when it was created doesn't become privileged just because it is later handed to counsel.

Work Product

The work-product doctrine protects materials prepared by or for a party's lawyer in anticipation of litigation. Heppner's lawyers acknowledged that he created the AI documents on his own, not at their direction. The court held that he was not acting as his lawyers' agent, and that documents which may later influence a lawyer's thinking are different from documents that reflect the lawyer's mental impressions when they are created. His AI documents fell into the first category, so they were not protected.

So, Should You Use AI to Communicate With Your Criminal Defense Attorney?

Unless your lawyer specifically advises you to use AI, you should NOT use AI to communicate with your attorney. The reasoning in Heppner points to some practical rules for anyone facing charges:

• Don't type the facts of your case into a public AI chatbot. That includes what happened, who was there, what you drank, what you said to police, and what you think the evidence shows. Assume it could be seen by someone other than you and your lawyer.

• Don't paste your lawyer's advice or emails into an AI tool. In Heppner, the fact that some inputs came from his lawyers did not protect them. Sharing your attorney's advice with a third-party platform may also put the confidentiality of that advice at risk.

• Don't upload discovery, police reports, or court documents to a chatbot. These often contain sensitive information about you and other people.

• Don't ask a chatbot what you should say or do about your case. It is not your lawyer, it doesn't know your record or your judge, and it can be confidently wrong about the law, sometimes even inventing cases and statutes.

• Remember that your devices and accounts can be searched. Heppner's AI documents were seized from his own devices under a search warrant. Chat histories stored on your phone, computer, or AI account can be reached through legal process.

• Ask your lawyer before using any AI tool for anything related to your case. The court noted the result might have been different had counsel directed Heppner to use the tool to prepare for their communications. That doesn't mean it would have been, and it depends on the tool, its terms, and the facts. But direction from your attorney matters, and it's a conversation to have before you start typing, not after.

How Far Does Heppner Go?

It's important not to overstate the case. Heppner is a federal trial-court decision from New York. It does not bind Pennsylvania state courts, though a Pennsylvania judge could find its reasoning persuasive. Several law firms and commentators have noted that the ruling turned on its specific facts: a consumer AI platform whose privacy policy allowed broad use and disclosure of user data, and a defendant who acted on his own without his lawyer's direction. Others have observed that the opinion's language is broad and could be read to reach client AI use more generally.

The law is also unsettled. On the very same day as the oral ruling in Heppner, a federal court in Michigan, in Warner v. Gilbarco, Inc., 2026 WL 373043 (E.D. Mich. Feb. 10, 2026), reached a different result in a civil case, holding that a self-represented plaintiff's ChatGPT queries and the responses she received were protected work product. Courts have since continued to grapple with the issue. Different tools, different privacy terms, different types of cases, and different levels of involvement by a lawyer can all lead to different outcomes.

Because of that uncertainty, the safest approach for anyone in a criminal case is to assume that a conversation with a public AI chatbot is not a confidential communication with your lawyer.

The Takeaway

AI tools can be useful for many things, but preparing your criminal defense on your own is not one of them. If you're facing a criminal, DUI, or traffic charge in Western Pennsylvania and have questions about what to write down, what to share, or how to prepare for your case, call the Zuckerman Law Firm at 412-447-5580 for a free consultation.