Using AI for Legal Advice Risks Waiving Attorney-Client Privilege
Court rulings warn AI-assisted legal advice can waive attorney-client privilege.
Why it matters: Clients and legal teams relying on AI tools risk exposing sensitive information without protection. Understanding privilege boundaries is crucial for general counsels and legal tech users to avoid inadvertent waiver.
- On Feb 10, 2026, Judge Jed S. Rakoff ruled AI-generated documents aren't protected by attorney-client privilege in United States v. Heppner.
- The defendant created about 31 documents with AI chatbot Claude before sharing them with counsel.
- The court noted AI tools’ data policies allow third-party access, undermining confidentiality.
- Some courts view AI as instruments—not third parties—so privilege impact varies across cases, e.g., Warner v. Gilbarco, Morgan v. V2X Inc.
In a key ruling on February 10, 2026, Judge Jed S. Rakoff of the U.S. District Court for the Southern District of New York found that documents generated by AI tools are not protected by attorney-client privilege. This decision arose in United States v. Heppner, where the defendant used the AI chatbot Claude to create roughly 31 defense-related documents before consulting his lawyers.
Judge Rakoff emphasized that interactions with AI are not communications with legal counsel, saying, "I'm not seeing remotely any basis for any claim of attorney-client privilege." Additionally, the court highlighted that AI platforms’ privacy policies often permit data collection and sharing with third parties, which destroys any reasonable expectation of confidentiality, a critical element for privilege protection.
However, the landscape remains unsettled. In cases like Warner v. Gilbarco, Inc. and Morgan v. V2X Inc., courts determined AI tools should be viewed as instruments rather than third parties. Consequently, use of AI did not automatically waive work product protections in these rulings.
These diverging views underscore the risks for legal professionals and their clients when leveraging AI for legal advice or document generation prior to official attorney engagement. Clients must exercise caution, understanding that early reliance on AI could jeopardize the confidentiality and privileged status of their communications.
Legal teams should establish clear protocols on AI tool usage and educate clients about the potential for privilege loss. Detailed scrutiny of AI platforms’ privacy policies and opting for secure, enterprise-grade AI services may offer better safeguards against accidental disclosure.
By the numbers:
- 31 — number of AI-generated documents in the Heppner case
- February 10, 2026 — date of the pivotal ruling by Judge Rakoff
- 2026 — year multiple courts issued divergent rulings on AI and privilege
Yes, but: Some courts distinguish AI tools as instruments, not third parties, potentially preserving privilege in some circumstances, but consensus is lacking.
What's next: Ongoing litigation and regulatory guidance are expected to clarify how attorney-client privilege applies in the context of AI-generated legal materials.