OPEN LINK ↗
// 45d agoNEWS
Claude ruling says chatbot isn't counsel
In United States v. Heppner, the SDNY held that documents a defendant created using Anthropic’s Claude were not protected by attorney-client privilege or the work-product doctrine. The court’s core point was straightforward: Claude is not an attorney, was not acting as counsel’s agent, and later sharing AI-generated material with lawyers does not retroactively make it privileged. The decision is really about the risks of using consumer AI for legal strategy, not a blanket rule that all AI-assisted legal work is public.
// ANALYSIS
Hot take: this is less a shocking new legal principle than a judge finally applying old privilege doctrine to a public chatbot.
- –A consumer LLM is a third party, so prompts and outputs can lose confidentiality fast.
- –Privilege is about confidential communications with counsel or counsel’s agents, not private brainstorming with a model.
- –The opinion leaves room for a different result if a lawyer specifically directs the AI use and can fit it into an agency-style argument.
- –The practical takeaway is simple: do not treat public AI tools like confidential legal channels.
// TAGS
claudeanthropicailegalprivilegeattorney-client-privilegework-productcourt-ruling
DISCOVERED
45d ago
2026-04-17
PUBLISHED
45d ago
2026-04-16
RELEVANCE
8/ 10
AUTHOR
ColdPlankton9273