Court Splits on Work-Product Protection for AI Chat Histories

A federal magistrate judge in the Southern District of Florida has ruled that a represented party's AI-assisted research can retain work-product protection — but only some of it. The decision in Fontainebleau Florida Hotel, LLC v. Botach, issued earlier this month, lands amid a fast-developing split over whether chatbot conversations are privileged, and it arrives as courts nationwide confront a problem practitioners increasingly face: clients are consulting AI tools without telling their lawyers.

What did the Fontainebleau court hold?

The plaintiff sought discovery of many of the defendant's communications with third-party AI providers. The defendant resisted, citing Warner v. Gilbarco, Inc., 820 F. Supp. 3d 629 (E.D. Mich. 2026), for the proposition that "litigation-related materials do not automatically lose work-product protection merely because a party used a third-party AI platform to prepare them."

The magistrate judge agreed that using AI tools does not necessarily waive work-product protection. He concluded the doctrine could shield some, but not all, of the defendant's AI-assisted research.

The court confronted a competing precedent. In United States v. Heppner, 820 F. Supp. 3d 292 (S.D.N.Y. 2026), a court held that research performed independently by a defendant — not at the behest of counsel — fell outside work-product protection. The Fontainebleau court acknowledged that holding but found it did not establish "that all materials independently prepared by a represented party are categorically unprotected."

Instead, the court grounded its analysis in the Eleventh Circuit's reading of Rule 26(b)(3)(A), which "confirms that work-product protection is not limited to materials prepared by counsel but extends to materials prepared by or for a party or its representative in anticipation of litigation."

Protection was not automatic. Applicability turned on "the purpose and circumstances of the particular communications." On that test, queries related to "producing public-facing statements or pursuing Defendant's independent activities, rather than preparing the defense of this litigation" fell outside privilege.

To balance disclosure against protection, the court ordered the defendant to "identify the withheld material and describe the nature of the asserted protection sufficiently to permit Plaintiff to assess the claim without revealing the protected information itself."

How are other courts treating AI use?

The cases divide along two lines: who used the AI, and why.

In Morgan v. V2X, Inc., 2026 U.S. Dist. LEXIS 67939 (D. Colo. 2026), a pro se plaintiff asserted work-product protection over his AI use. The court held that even though publicly available AI tools often rely on user input to train their models, the plaintiff held a reasonable expectation of privacy sufficient to support the protection.

In Assini v. Hayward, 2026 N.Y. Slip Op. 26086, New York Supreme Court Justice Rhonda E. Fischer examined a pro se defendant's AI use. She concluded that because the responses were obtained in preparation for litigation, and because the defendant had a reasonable expectation of privacy and confidentiality, the work-product doctrine applied.

A separate question remains unresolved nationally: whether routing communications through a third-party AI platform itself waives protection. Some courts treat AI chats as disclosure to a third party. Others view them as tools no different from other technologies.

The status of the user matters too. Courts weigh whether a litigant is pro se, whether represented parties conducted research at their attorney's request, or whether they asked a third party to query the AI tool on their behalf.

What does this mean for practitioners?

Content is king, but so is context. Counsel should ask two questions when AI use surfaces in discovery: who was using the tool, and why were they using it.

The practical takeaway is preventive. Lawyers should raise AI use at intake, asking whether clients have already discussed the dispute with a chatbot and what they shared. Those answers can flag discovery exposure before it derails a case. Counsel should also explain when clients must notify them about chatbot interactions concerning the litigation. Clients will use these tools regardless — and a few casual queries, disclosed late, can reshape the discovery fight.

As the split deepens, transparency at the start of a case may make all the difference.