Courts Clash on AI Privilege as Heppner and Warner Rulings Diverge

Got story updates? Submit your updates here. ›

Two recent federal court rulings, Heppner and Warner, reached different conclusions on whether the use of AI tools like ChatGPT and Anthropic's Claude can destroy attorney-client privilege and work product protection. While the outcomes may both be defensible on their specific facts, the courts adopted incompatible frameworks for how AI tools relate to these foundational legal doctrines. This jurisdictional divide leaves practitioners uncertain about the risks of using AI in litigation and highlights the need for clearer guidance from higher courts.

Why it matters

The conflicting Heppner and Warner rulings create a jurisdictional nightmare for lawyers, as the use of consumer AI tools could be deemed to waive privilege in some courts but not others. This uncertainty threatens to undermine the confidentiality that is essential for effective legal representation. The stakes are high, as AI is becoming increasingly prevalent in legal workflows, from document drafting to legal research. Resolving this doctrinal split is crucial for providing clear guidance to the legal profession on how to navigate the privilege implications of AI use.

The details

In Heppner, Judge Jed Rakoff ruled that the use of Anthropic's consumer version of Claude destroyed attorney-client privilege and work product protection, citing the tool's privacy terms that allow for training on user data and potential disclosure to regulators. In contrast, in Warner, Magistrate Judge Anthony Patti held that a pro se litigant's use of ChatGPT was protected work product, reasoning that AI systems are tools, not third parties, and that disclosure to a tool does not constitute waiver. The key divergence between the two rulings is in how the courts view AI tools in relation to privilege and work product doctrine. Heppner treats AI as a third party that can defeat confidentiality, while Warner sees AI as a tool that does not "receive" legal strategy in the same way a human consultant would. This fundamental disagreement on the nature of AI systems is at the heart of the incompatible frameworks adopted by the courts.

• On February 10, 2026, Judge Rakoff ruled from the bench in Heppner, later issuing his written opinion on February 17.
• On that same February 17, Magistrate Judge Patti held in Warner v. Gilbarco, Inc. that a pro se litigant's ChatGPT-assisted drafting was protected work product.

What they’re saying

What’s next

The conflicting Heppner and Warner rulings are likely to be appealed, and higher courts will eventually have to choose between the incompatible frameworks adopted by the district and magistrate judges. This issue is well-suited for Supreme Court review, as the courts' divergent approaches to how AI tools relate to privilege and work product doctrine will have far-reaching implications for the legal profession.