Source: JD Supra

JD Supra explores the diverging legal paths taken by federal courts in February 2026 regarding the discoverability of AI interactions. In US v. Heppner (SDNY), Judge Rakoff held that AI chats are not privileged because the tool is a third party with no duty of confidentiality. However, on the same day in Warner v. Gilbarco (ED Michigan), Magistrate Judge Anthony Patti ruled that a pro se litigant’s queries to ChatGPT were protected by the work-product doctrine. Judge Patti reasoned that the AI queries reflected the plaintiff’s internal mental impressions and that disclosure to the software was not "disclosure to an adversary."

The contrast between these rulings reveals a developing "fact-specific" framework for AI law. While Rakoff focused on the consumer-grade privacy policies and the lack of attorney supervision, Patti emphasized that AI is a tool akin to an "electronic brainstorming session" for a litigant preparing for trial. This split creates significant uncertainty for corporate litigants who must now decide whether to claim work-product protection for their internal AI prompts. Experts advise that until higher courts resolve the conflict, the safest course is to ensure all AI legal tasks are formally commissioned by counsel to strengthen the "anticipation of litigation" argument.