By Lansing Family Law Attorneys

What Happened in U.S. v. Heppner

In United States v. Heppner, No. 25 Cr. 503 (S.D.N.Y. Feb. 17, 2026), a CEO facing federal securities fraud charges made a costly mistake. He opened the free consumer version of Claude — an AI chatbot made by Anthropic — and started typing in legal strategy. Information his own lawyers had shared with him went straight into those prompts.

Prosecutors asked for those AI-generated documents. The court said: hand them over.

No privilege. No work-product protection.

The reasoning was straightforward. Claude is not an attorney. Anthropic’s consumer terms do not promise confidentiality. And the defendant acted entirely on his own — no attorney directed him to use the tool.

This was a federal case out of the Southern District of New York. Federal courts routinely look to decisions from other circuits and districts when resolving privilege disputes. Michigan state courts may do the same.

Why this isn’t Considered a Breach of Privacy

An average person would see this as a breach of privacy, but the Court does not see it that way. Once Heppner typed that information into Anthropic’s consumer platform, he voluntarily shared it with a third party. Under the third-party doctrine, privacy protections generally evaporate when you willingly hand information to someone else. Anthropic’s terms of service made clear they could retain and access those inputs. So legally, there was no longer a “private” communication to protect. 

The reason this is different from the Court viewing your search history is because applications like Google do not say they will share your searches with others in their terms of service while Anthropic does. Heppner was trying to expand on the advice given to him by his lawyer — something people do every day by Googling, reading articles, or even by asking AI. The idea that doing so privately could become evidence against you is extremely troubling. If he had just Googled the same questions, that search history would have been far harder for prosecutors to obtain. Using AI, which felt more private and helpful, actually made it easier to prosecute him. 

Many people assume using AI feels similar to privately Googling legal questions. But modern search engines increasingly filter your searches into their own generative AI. Google’s Gemini-powered summaries, for example, may process your searches differently than a traditional search. If you are researching a sensitive legal issue — divorce, criminal allegations, CPS investigations, or financial disputes — consider limiting the amount of identifying information you enter into AI-powered systems. Some users also choose to disable AI-generated search features, turn off Web & App Activity tracking, use private browsing modes, or use privacy-focused search engines to reduce data retention and third-party exposure. While no online activity is perfectly private, minimizing unnecessary disclosure can help reduce confidentiality risks. 

Why This Matters for your Privacy

If you type detailed information about your divorce, criminal case, or custody dispute into a consumer AI chatbot, that information may no longer remain fully private, and you lose the confidentiality you have with your attorney.

The short version: if you type sensitive information into the wrong AI platform, that information may not remain private and could potentially be used against you later. 

Why Free AI Tools Carry More Risk

Some business and law-firm AI systems include stronger confidentiality protections than free public AI tools. But most consumers use the public versions of ChatGPT, Claude, Gemini, and similar systems. Those platforms may retain, review, or process what users type into them. 

Practical Ways to Protect Your Privacy When Using AI 

Think Carefully About What You Share 

Before typing into an AI chatbot, ask yourself whether you would feel comfortable seeing that information repeated in court. Divorce details, custody disputes, medical records, financial documents, and criminal allegations may become far less private once entered into a consumer AI platform. 

Read the platform’s terms.

Not a summary. The actual terms of service. Before using an AI platform for sensitive questions, take a few minutes to understand how the company handles your information. Some platforms store conversations, allow human review, or use user data to improve their systems. 

Simple Ways to Protect Yourself

• Avoid typing full names, addresses, Social Security numbers, or case numbers into AI tools.
• Do not upload court filings, medical records, or financial statements into free AI chatbots.
• Use AI for general education, not personalized legal strategy.
• Avoid asking AI whether you should hide assets, delete evidence, or violate court orders.
• If you already entered sensitive information into an AI platform, tell your attorney early.

Common Questions

Does using AI automatically waive attorney-client privilege?

Not automatically. But using consumer-grade AI tools — the free versions of ChatGPT, Claude, Gemini, and similar products — creates a serious risk of waiver, as Heppner illustrates. The platform’s terms of service matter enormously.

Can AI Conversations Be Used in Court? 

Potentially, yes. As Heppner illustrates, information shared with AI platforms may not receive the same privacy protections as conversations with your attorney. Courts may view information voluntarily entered into AI systems as shared with a third party. 

The Bottom Line

AI is not going away. Neither is the risk it creates when used carelessly.

AI tools can be useful for learning about legal issues, but they are not always private. Before typing sensitive information into an AI chatbot, remember that what feels like a private conversation may not legally remain one. The safest approach is to avoid sharing identifying or confidential information online until you have spoken directly with an attorney you trust. 

People often treat AI chatbots like private conversations, but courts may not see them that way. Being careful about what you share online can help protect your privacy during a legal dispute. 

At The Kronzek Firm, we proudly serve clients throughout Mid-Michigan, including Ingham, Eaton, Clinton, and Livingston Counties. Whether you’re navigating a divorce, criminal investigation, or CPS case in Lansing, Mason, St. Johns, or Howell, speaking directly with an attorney is far safer than relying on AI-generated advice alone. Contact us today for a consultation at contactus@kronzek.law.

📞 For more than 30 years, our attorneys have fought for thousands of Michigan clients. Reach our office at (517) 886-1000.