Fontainebleau Florida Hotel, LLC v. Botach, No. 25-cv-20251 (S.D. Fla. Oct. 21, 2026), full opinion (PDF)
A federal magistrate judge in Miami has allowed discovery of a defamation defendant's nonprivileged communications with AI platforms about the hotel suing him. The ruling takes up when a litigant's prompts to AI platforms are discoverable. It also addresses how far work-product protection reaches over those prompts. The hotel sought the communications through a request for production and through proposed subpoenas to four AI service providers.
The court rejected the defendant's blanket work-product objection to the communications it allowed. It agreed with him that requests about the litigation in general, the court's orders, and the judges fell outside the discovery the district judge had authorized. The court also denied the hotel's request to compel the defendant's consent to the provider subpoenas.
What happened
The district judge had reopened discovery limited to damages. The hotel seeks punitive damages without Florida's statutory cap. Its theory depends in part on showing that the defendant specifically intended to harm the hotel. In the hotel's view, the defendant's prompts and revisions could show his intent and his efforts to amplify the harm.
The defendant objected that the requests exceeded the purpose for which discovery was reopened. He also argued that they risked exposing work product, including AI-assisted litigation preparation and legal research.
The court's analysis
The court concluded that communications about the hotel, its counsel, and its employees, owners, or representatives fall within the authorized damages discovery. If the defendant used AI to generate or revise statements about those subjects, the court reasoned, the communications "may reveal not merely the final content of Defendant's statements but also the purpose for which the statements were created".
Requests about the litigation as a whole, the orders entered in it, and the judges lacked a sufficient connection to damages, in the court's view. The court noted that a communication about the hotel or its people remains subject to production even when it also mentions the litigation, an order, or a judge. The exception is a communication covered by a valid privilege or protection.
The court rejected the blanket work-product objection. It stated that it "must examine the purpose and circumstances of the particular communications rather than presume protection from the medium through which they were created". Communications the defendant generated on his own to create or revise public-facing statements, the court reasoned, ordinarily fall outside the protection.
The court explained that a protected document must have been "created principally or exclusively to assist in anticipated or ongoing litigation". The party asserting protection bears the burden of proving that purpose. Under the order, the defendant remains free to withhold a particular communication that meets the Rule 26(b)(3) standard. Each withheld communication must appear on a privilege log under Rule 26(b)(5)(A).
The hotel had also asked the court to compel the defendant's consent to the provider subpoenas. It cited concern that he may have deleted or failed to preserve responsive communications. The court denied that request. It explained that Rule 34 requires a party to produce responsive material within its possession, custody, or control. In the court's view, the rule does not expressly require a party to authorize a third party to disclose records to an adversary.
The hotel may serve Rule 45 subpoenas limited to the permitted subjects. The court left room for other relief on a particularized showing of a specific impediment that ordinary subpoena procedures cannot address. If the providers produced directly to the hotel's counsel, the court reasoned, the defendant would lose the chance to screen the production for work product. Before the subpoenas issue, the parties must meet and confer on a procedure for the providers' production.
Why it matters
When intent is at issue, a litigant should expect its AI communications about the opposing party to be discoverable on the ordinary relevance showing. Counsel should identify the AI platforms a client has used to write about the opposing party and bring those accounts within the litigation hold.
Work-product claims over AI prompts must be supported item by item. Before withholding any prompt, counsel should be able to show that the client created it principally to assist in the litigation. Building the privilege log entry by entry as the review proceeds keeps that showing ready.
When drafting requests for AI communications, tie each category to a claim or a damages element rather than to the litigation itself. Requests framed around the lawsuit, its orders, or its judges are the most exposed to a scope objection, especially while discovery is limited.
Before issuing provider subpoenas, a requesting party should be ready to agree on a production procedure that accounts for work-product objections. It should plan on Rule 45 rather than a compelled consent.
The full opinion is available as a PDF.
