Public adjusters have good reasons to be interested in artificial intelligence (“AI”). A tool that helps organize a complicated claim, compare estimates, identify missing documentation, and improve a letter can leave more time for the work that requires experience and judgment. I understand the attraction. I also think a recent Florida federal court decision deserves a place in our conversations about how to use these tools well.

In our litigation, we are now seeking information about insurers’ and their consultants’ use of AI in the claims process. We want to understand what information goes into these systems, what instructions guide them, and how their output affects decisions about our clients’ claims. Discovery runs both ways. Public adjusters should expect insurance defense counsel to pursue the same questions about their work.

You can bet an insurance company lawyer would be interested in finding a prompt that says, “Make the estimate higher,” or “Write a letter explaining why the insurance adjuster is acting in bad faith so I can leverage a higher amount in settlement negotiations.” Those are hypothetical examples, but the discovery issue is real. The finished estimate or letter may be only part of what the other side asks to examine.

A short prompt does not necessarily tell the whole story. An estimate may need to be higher because covered damage was omitted, quantities were wrong, or necessary repair work was overlooked. An insurer’s conduct may warrant a firm letter and an investigation into bad faith. Policyholders are entitled to vigorous advocacy. The concern arises when a requested result takes the place of the factual analysis needed to support it. An isolated prompt can also invite an argument about intent and motivation that the fuller record does not support.

In Fontainebleau Florida Hotel, LLC v. Shmuley Boteach Jacob, a federal magistrate addressed demands for a defendant’s AI communications in a defamation lawsuit. 1 Discovery had reopened for the limited purpose of determining damages. The hotel sought communications about itself, its personnel, its lawyers, the lawsuit, court orders, and the judges.

The court permitted targeted discovery concerning the hotel, its personnel, and its counsel, subject to valid protections. The significance extended beyond the words ultimately published. AI prompts and revisions could reveal the purpose behind statements and the effects their author sought to achieve. The court did not decide that those exchanges proved improper intent. It found a sufficient connection to the disputed damages issues to permit discovery of these AI chats and exchanges.

For public adjusters, the practical implication is that the process of creating a document may receive as much attention as the document itself. An insurer’s lawyer may want to know whether an estimate grew because additional damage was identified or because someone instructed a program to reach a larger number. We would ask comparable questions if an insurer instructed its system to reduce an estimate or find arguable reasons to deny a claim.

I encourage thoughtful and honest use of AI. A useful assignment might ask the system to compare an estimate with inspection notes, identify potentially omitted work, cite the supporting material, and flag uncertainty. The adjuster must then check the results. That approach helps produce a more complete claim and a record explaining why it is more complete. Dressing up a predetermined answer in careful language accomplishes neither.

There is another reason for taking care when using AI. The law governing protection of AI exchanges is unsettled. A recent Fintech Law Report article, “When a Non-Lawyer Officer or Director Asks an AI Platform for Legal Advice on a Company Matter, Are the Conversations Protected from Discovery in Litigation?—Fortis v. Krafton,” 2 makes the point directly: “The law on this issue is unsettled and evolving.” I think that warning is significant. This is new to lawyers and to those using AI whose work is questioned in a lawsuit.

The decisions discussed in that article reach different outcomes under different circumstances. Some recognize protection for AI-assisted litigation preparation. Others reject claims of protection. Relevant questions include the purpose of the work, counsel’s involvement, the platform’s terms, and the applicable procedural rules. Favorable decisions involving people representing themselves should not be treated as assurances that a public adjuster’s exchanges will receive the same treatment.

Fontainebleau recognizes that work-product protection can extend beyond materials created by attorneys. A public adjuster may therefore have a legitimate protection claim for qualifying work prepared for anticipated litigation on the policyholder’s behalf. But an ordinary adjusting task does not become protected simply because a lawsuit later follows. Adding an attorney to an email or marking a document “privileged” does not explain why the work was created.

My practical advice is to involve counsel when the claim first looks like it may need litigation to resolve. Genuine litigation preparation, documented for that purpose by counsel, and understanding the confidentiality arrangements for the technology being used are going to become more important. Keep routine claim work and confidential litigation analysis organized so their purposes can be explained. Even where the substance of an exchange is protected, questions about the tool used or the documents uploaded may remain discoverable.

Preservation matters as well. When litigation is reasonably anticipated, relevant AI records should be addressed with counsel along with emails, photographs, estimates, and other claim materials. That includes considering prompts, responses, uploads, and automatic deletion settings. Deleting an uncomfortable conversation can create a much more serious problem than explaining it.

The public adjusters I know take pride in finding damage others missed and explaining what a proper repair requires. AI can help with that work. It can also generate confident language that outruns the evidence. The professional reviewing the result still needs to verify the facts, quantities, policy language, and conclusions before relying on them.

I would encourage public adjusters to approach this new technology with the same good faith and honest intent they bring to an inspection or a negotiation. Ask it to help investigate, test, organize, and explain. Be willing to learn that an assumption was wrong. A well-supported claim can withstand hard questions, including questions about how AI helped prepare it. Policyholders deserve both the benefits of better tools and the judgment to use them well.

Thought for the Day

“This new generation of AI will remove the drudgery of work and unleash creativity.”
—Satya Nadella


1 Fontainebleau Florida Hotel v. Jacob, No. 25-CV-20251 (S.D. Fla. Oct. 21, 2026).

2 Gail Weinstein, et al. “When a Non-Lawyer Officer or Director Asks an AI Platform for Legal Advice on a Company Matter, Are the Conversations Protected from Discovery in Litigation?—Fortis v. Krafton,” Vol. 29, No. 4 Fintech L. Rep. NL 1 (July/August 2026).