Courts just ruled your ChatGPT chats are protected
Legal professionals already advise their clients to set up separate accounts for AI litigation work to ensure confidentiality
Litigators now face a question that didn't exist five years ago: are the prompts a party types into ChatGPT while building a case discoverable by the other side? Within two days this June, courts in Texas and New York answered for the first time, and both sided with protection.
In a decision issued on June 3, 2026 in Tate Group Automotive, LLC v. Legacy Automotive Capital (LLC), Judge Grant Dorfman of the Texas Business Court, Eleventh Division, decided that ChatGPT chats of Kris Tate, who is the principal of the business but not an attorney, fell under the attorney work-product privilege.
Are your ChatGPT litigation chats discoverable?
In making this determination, the judge cited Texas Rule of Civil Procedure 192.5, which provides protection for materials generated by or for a party rather than just by attorneys.
Most importantly, however, Dorfman required Tate Group to reveal which of the discovery materials had been inputted into ChatGPT.
The next day, Justice Rhonda Fischer of the Supreme Court, Nassau County, quashed a subpoena in Assini v. Hayward seeking a self-represented defendant's entire OpenAI account.
The court held that using a commercial AI tool doesn't itself waive confidentiality, comparing third-party technical access to cloud storage. Both rulings rejected the argument that entering case strategy into ChatGPT hands it to the opposing side.
In Conservation Law Foundation vs Shell Oil Co., Magistrate Judge Thomas Farrish ordered climate historian Dr Naomi Oreskes to produce the AI prompts she used to filter Shell's document production while preparing her expert report.
The discovery of an expert’s methodology is allowed by Farrish because, unlike an individual’s own strategy of litigation, it is discoverable under Rule 26 in order for the opposing side to verify the accuracy of the opinion. The order in question is stayed pending objection.
The decisions have created a line: the use of AI for the development of legal strategies is protected, while the use of AI for the production of evidence or expert opinions is discoverable.
Legal professionals already advise their clients to set up separate accounts for AI litigation work to ensure confidentiality.
-
Sony and Microsoft seeking to dismiss consumer lawsuits demanding tariff refunds
-
UK retailer John Lewis sees 8-fold rise in AI shopping
-
Nvidia to acquire Hugging Face in nearly $13 billion deal: Here’s why
-
AI agents can hide memory attacks until it’s too late
-
WhatsApp to remind you about unsent messages: Report
-
Apple hit with £2bn UK lawsuit over app tracking rules
-
Americans are revealing their AI secrets: Here’s what they’re saying
-
Google Vids can now turn documents into videos: Here’s how
-
20 million children face online sexual abuse annually, UNICEF report reveals
-
Anthropic back in the Trump’s administration’s good books, US Commerce Secretary says
-
Nvidia's Jensen Huang urges G20 nations to avoid AI rules based on ‘hypothetical’ harms
-
Apple Maps renames Lake Ontario after Trump's order