Federal Court Rules AI-Generated Materials Lack Attorney-Client Privilege Protection
A U.S. District Court judge in New York has held that documents created using consumer AI tools like ChatGPT and Claude are not protected by attorney-client privilege or work product doctrine, meaning they can be discovered and disclosed in legal proceedings.
The ruling emerged from United States v. Heppner, a securities fraud case where the defendant used Claude to analyze his legal situation after receiving a grand jury subpoena. He later shared the AI-generated summaries with his lawyer and claimed privilege protection. The court rejected that claim.
Why the Court Denied Privilege Protection
The judge identified four critical problems with Heppner's privilege claim:
- Communications with Claude were not communications with his attorney
- The materials were created independently, not at counsel's direction or suggestion
- Consumer AI platforms lack confidentiality protections-Anthropic's privacy policy allows data collection for training and disclosure to third parties, including government authorities
- Sharing unprivileged documents with a lawyer after creation does not retroactively make them privileged
The work product doctrine also failed to protect the materials because Heppner created them on his own initiative, not by or at the direction of counsel.
What This Means for Your Practice
While this is a single criminal case, civil litigators should expect the same reasoning to apply in discovery disputes. Opposing counsel can now request your ChatGPT prompts and outputs if they relate to matters where litigation is anticipated or underway.
The practical risks are substantial. Information you input into consumer AI tools may become evidence against you. Early thinking, legal analysis, business strategy, and financial data shared with these platforms can surface in discovery.
What You Should Do Now
Do not input sensitive information into consumer AI tools. Avoid uploading contracts, legal documents your counsel provides, questions tied to active matters, or business strategies.
If your firm uses AI tools, restrict use to enterprise platforms that maintain strict confidentiality protections and do not train on your data. Consult counsel before using any AI for legal or regulatory questions tied to matters where litigation is reasonably anticipated.
The law in this area remains unsettled. Courts may eventually distinguish between consumer platforms and enterprise tools designed to preserve confidentiality. Until clearer guidance develops, assume that consumer AI communications are not confidential.
Learn more about AI for Legal professionals and how to safely incorporate these tools into your practice.
Your membership also unlocks:
