Be careful: chatting with AI about your case is discoverable

a day ago

Court in United States v. Heppner ruled that communications between a defendant and generative AI platform Claude were not protected by attorney-client privilege or work product doctrine.
The decision categorically excludes a client's self-directed use of AI from privilege, but a more fact-dependent analysis could suggest it should sometimes qualify.
The court's reasoning focused on Claude not being an attorney, lack of confidentiality due to third-party AI platform policies, and the purpose not being for legal advice from Claude.
Argument is made that AI should be treated as a tool like Google Docs or Gmail, not as a third-party human, to align with existing privilege treatments of technology.
Confidentiality analysis should center on a client's reasonable expectation of privacy, considering opt-out options and technical data handling, rather than blanket policies.
The purpose test should ask if AI was used to facilitate obtaining legal advice from an attorney, similar to self-directed notetaking cases, not if advice was sought from the AI itself.
The ruling may discourage client collaboration and innovation in attorney-client relationships, and future courts should adopt a functional, case-by-case approach to AI and privilege.

Hasty Briefsbeta