Court: AI chats not privileged
A federal court ruled that communications with generative AI tools are not protected by privilege, creating fresh compliance and evidence‑handling headaches for insurers using AI in claims, underwriting or investigations. The decision raises auditability and disclosure risk for AI‑driven workflows. (reuters.com)
The matter is United States v. Heppner, No. 25 Cr. 503 (S.D.N.Y.), where Judge Jed S. Rakoff issued a bench ruling on Feb. 10, 2026 and followed with a written opinion on Feb. 17, 2026 after defendant Bradley Heppner used the consumer version of Anthropic’s Claude. (debevoise.com) Federal agents seized devices from Heppner’s home on Nov. 4, 2025 and recovered approximately 31 documents memorializing exchanges between Heppner and Claude that the government later moved to inspect. (debevoise.com) Judge Rakoff found the AI outputs were not privileged because (1) Claude is not an attorney and thus not a communication with counsel, (2) the platform’s terms and lack of confidentiality undermined any reasonable expectation of secrecy, and (3) the materials were not created at counsel’s direction so did not qualify as work product. (crowell.com) The court ordered production of the seized AI-generated documents to the government, treating them like third‑party materials rather than protected client communications. (paulweiss.com) Post‑ruling guidance from major firms has urged that privilege may only be preserved where counsel expressly directs AI use, that firms should document that direction, and that organizations implement formal AI‑use protocols to limit disclosure risk. (goodwinlaw.com) Industry observers report insurers and financial institutions are already reassessing employee use of public generative AI tools because the Heppner record — and the government’s access to the 31 Claude documents — creates concrete auditability and disclosure exposure for claims, SIU, and underwriting workflows. (insurancebusinessmag.com)
