Judge Rakoff Rules 31 Claude Chats Not Privileged in $150 Million Fraud Case
Updated
Updated · Forbes · May 28
Judge Rakoff Rules 31 Claude Chats Not Privileged in $150 Million Fraud Case
3 articles · Updated · Forbes · May 28
A Feb. 17 order by U.S. District Judge Jed Rakoff let prosecutors use 31 saved Claude conversations in which Bradley Heppner sought a criminal defense strategy after learning he was under investigation.
Rakoff said the chats were not protected by attorney-client privilege or work-product doctrine because Claude is not a lawyer, cannot form the required human confidential relationship, and Anthropic’s terms allow disclosure of user data.
The judge said a narrow exception might exist if a lawyer had directed a client to use AI as the lawyer’s agent, but Heppner acted on his own and his attorney learned of the chats only later.
Heppner, former CEO of GWG Holdings, was convicted this month in a securities-fraud case alleging more than $150 million in investor losses and now faces up to 20 years in prison.
Rakoff called the issue a nationwide first impression, and lawyers said the ruling warns executives that chats with Claude, ChatGPT or Gemini can be discoverable in both criminal and civil cases.
With courts divided on AI's legal status, are your private chatbot conversations actually safe from discovery?
A businessman used AI to plan his legal defense. Why did it become the prosecution's star witness?
Landmark 2026 Federal Ruling: AI-Generated Content Not Protected by Legal Privilege—Implications and Safeguards for Organizations
Overview
In March 2026, Judge Jed S. Rakoff issued a landmark federal ruling in the Heppner case, deciding that AI-generated content created with a publicly available, non-confidential tool and without direct attorney supervision is not protected by legal privilege or work product doctrine. The court examined documents made by Bradley Heppner using such an AI tool, and found that these specific facts were central to denying privilege protection. This decision sets an important precedent, highlighting that the use of public AI tools without legal oversight can expose sensitive information and complicate legal proceedings.