A federal magistrate judge in Florida has ruled that a person who uses artificial intelligence to prepare for a lawsuit may be entitled to work-product protection for certain AI prompts and outputs—even if the person created those materials independently, without an attorney’s instructions.
The October 21, 2026, ruling in Fontainebleau Florida Hotel, LLC v. Botach, No. 1:25-cv-20251 (S.D. Fla.), addresses an increasingly important discovery question: When can an opposing party obtain conversations that a litigant had with a third-party AI platform? U.S. Magistrate Judge Marty Fulgueira Elfenbein concluded that the answer depends on why each communication was created, not whether an attorney personally directed the exchange.
Work-Product Protection Is Not Reserved for Attorneys
Federal Rule of Civil Procedure 26(b)(3)(A) generally protects documents and tangible things prepared in anticipation of litigation or for trial “by or for another party or its representative.” The court emphasized that this language is not confined to attorney-created materials. Drawing on Eleventh Circuit precedent, including Tambourine Comercio Internacional SA v. Solowsky and Republic of Ecuador v. Hinchee, Judge Elfenbein explained that qualifying materials prepared by a party may receive protection even when no lawyer created or commissioned them.
That distinction is significant for people who independently use generative AI to research legal issues, organize evidence, analyze potential claims, or explore litigation strategies. An AI conversation is not necessarily stripped of work-product protection solely because a client initiated it on a personal account rather than at counsel’s request.
But the ruling does not create automatic protection for everything a litigant asks an AI chatbot. Under the standard applied by the court, the primary motivating purpose of creating the material must have been to assist anticipated or ongoing litigation. Merely generating a document after a lawsuit was filed—or later finding it useful in the case—is insufficient. The person asserting protection must substantiate the claim.
How the Florida Ruling Addresses Heppner and Warner
The court considered two other 2026 decisions concerning AI-assisted legal work. In United States v. Heppner, a federal court in New York found that certain AI communications independently created by a represented defendant were not work product where they were not directed by counsel and did not reflect counsel’s strategy. Judge Elfenbein agreed that being represented by an attorney does not automatically protect self-directed AI conversations. She declined, however, to read Heppner as categorically excluding materials independently prepared by a represented litigant.
The court also discussed Warner v. Gilbarco, Inc., a Michigan federal decision recognizing that AI-assisted litigation preparation can qualify as work product and that using a third-party AI tool does not necessarily waive protection. Judge Elfenbein explained that Warner likewise did not justify shielding every AI exchange about a dispute. Each communication must be evaluated on its own facts.
The Hotel’s Discovery Requests—and Their Limits
The issue arose in a defamation and business-interference lawsuit brought by the owner of the Fontainebleau Miami Beach against Jacob Shmuel Botach, also known as Shmuley Boteach. The hotel alleges that he published false statements and AI-generated videos about the hotel and people associated with it. Those are allegations in the complaint, not findings made in the October 21 discovery ruling.
After liability was resolved by default as a sanction, discovery continued on damages. The hotel sought AI communications relating to the hotel, its representatives and attorneys, as well as broader subjects such as the litigation and the judges. Judge Elfenbein permitted specified categories of communications potentially relevant to intent and punitive damages while rejecting other categories as beyond the authorized damages discovery.
Critically, the judge overruled the defendant’s blanket work-product objection but expressly preserved the opportunity to withhold qualifying individual communications. Any such claim must be properly described on a privilege log under Rule 26(b)(5)(A). AI prompts used to develop public-facing statements or social media posts ordinarily would not qualify merely because they concern the same underlying dispute.
What the Ruling Means for Litigants Using AI
The decision offers an important but limited protection. A litigant’s independent AI-assisted case preparation can fall within the work-product doctrine, even without counsel’s direction. At the same time, AI chats can become relevant discovery evidence, especially when prompts reveal a user’s objectives, revisions, or intent in creating material later made public.
Anyone using AI in connection with a potential or pending lawsuit should distinguish litigation-preparation materials from ordinary research, business communications, and public messaging; preserve relevant records; and consult counsel before disclosing sensitive case information to a third-party service. Whether a particular exchange is protected—and whether protection may be waived—requires a separate legal analysis. The October 21 order does not resolve every question about AI-platform confidentiality or waiver.
The ruling is a magistrate judge’s case-specific discovery order, not a nationwide rule. Nevertheless, it reinforces a central principle: protection turns on the purpose and circumstances of the work, not simply on whether a lawyer pressed the keys.
Read the full order and complaint: Fontainebleau Florida Hotel, LLC v. Botach, No. 1:25-cv-20251 — S.D. Fla. Second Order After Discovery Hearing (Oct. 21, 2026) and Fontainebleau Florida Hotel, LLC v. Botach — Complaint, Miami-Dade Circuit Court (Dec. 19, 2024).


