United States v. Heppner
Summary
In *United States v. Heppner*, Judge Rakoff of the Southern District of New York ruled that communications between a criminal defendant, Bradley Heppner, and the generative AI platform Claude were not protected by attorney-client privilege or the work product doctrine. This decision, described as a "question of first impression nationwide," stemmed from Heppner's use of Claude, developed by Anthropic, to prepare defense strategies and legal arguments after receiving a grand jury subpoena in October 2025. The court found that Claude was not an attorney, that Heppner lacked a reasonable expectation of confidentiality due to Anthropic's privacy policy allowing data collection for training and disclosure to third parties, and that Heppner did not use Claude for the purpose of obtaining legal advice *from Claude*. The analysis suggests the court's opinion leans towards categorically excluding client-initiated generative AI use from privilege, contrasting with a proposed fact-dependent approach that might sometimes qualify such use.
Key takeaway
For legal professionals advising clients on technology use, the *Heppner* ruling underscores the critical need to establish clear guidelines for generative AI interactions. You should explicitly direct clients on how to use AI tools like Claude for legal preparation, ensuring such use is framed as facilitating communication with counsel, not as seeking advice from the AI itself. Additionally, review AI platform privacy policies with clients and advise them to opt out of data collection for training to bolster claims of confidentiality, as the court's categorical exclusion could otherwise expose sensitive information.
Key insights
Client-AI communications are not inherently privileged, but a fact-dependent analysis is crucial.
Principles
- Attorney-client privilege requires communication with an attorney.
- Confidentiality hinges on a reasonable expectation of privacy.
- Privilege aims to foster full and frank communication.
Method
Courts should apply a fact-dependent inquiry focusing on the client's reasonable expectation of confidentiality and whether AI use facilitates obtaining legal advice from counsel, rather than from the AI itself.
In practice
- Clients should opt out of AI data collection for training.
- Counsel should direct client AI use to establish privilege.
- Treat AI as a tool, not a human agent, for privilege analysis.
Topics
- Generative AI
- Attorney-Client Privilege
- Work Product Doctrine
- Legal Technology
- Confidentiality
Best for: CTO, VP of Engineering/Data, Director of AI/ML, Legal Professional, Policy Maker, AI Ethicist
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by Law and Technology - Harvard Law Review.