Overview
- A Southern District of New York decision in United States v. Heppner found that documents a defendant created with Anthropic’s Claude and later sent to his lawyers were not covered by attorney‑client privilege or work‑product protection.
- The judge said the AI tool is not an attorney, the platform’s terms stated user inputs were not confidential, and the materials were created by the client outside any request from counsel.
- Legal advisories now warn that chats with public tools about divorces, custody disputes, or workplace claims can be requested in discovery, including prompts, inputs, and outputs.
- Paying for a premium AI plan does not solve confidentiality if the provider’s terms allow storing, training on, or sharing user data, which can count as disclosure to a third party and waive privilege.
- Practitioners urge clients to keep sensitive facts out of public chatbots, to route any AI use through attorney direction, and to favor enterprise tools that do not retain or reuse data.