In a February 17, 2026, Memorandum Opinion, a federal district court judge in the Southern District of New York found that a criminal defendant’s communications with the generative AI platform, “Claude,” were not protected from discovery by either the attorney-client privilege or the work product privilege.  See, U.S. v. Heppner, Case 1:25-cr-00503-JSR (S.D.N.Y., Feb. 17, 2026). While the court’s analysis was based on federal (and not Texas) law as it pertains to when such privileges apply, all practitioners should advise their clients to exercise caution with their use of AI related to legal matters, paying special attention to this ruling as well as any further developments. 

In Heppner, criminal defendant Heppner (a financial executive from Texas) was indicted for securities fraud, wire fraud, conspiracy, making false statements to auditors, and falsifying corporate records. In general terms, Heppner’s charges were based on the allegation that he defrauded investors out of more than $150 million in connection with a financial services company he founded and controlled.  After Heppner received a grand jury subpoena, after he engaged counsel, and after it was clear he was the target of the government’s investigation, he used Claude…