Texas Business Court Weighs In On Discoverability of AI Prompts

The Texas Business Court has entered the growing national debate about whether conversations with AI tools like ChatGPT are discoverable, and it came down on the side of protection. In a minute entry filed June 3, 2026, Judge Grant Dorfman of the Eleventh Division ruled that a non-lawyer’s ChatGPT conversations, prepared in anticipation of litigation, can qualify as protected attorney work product under Texas procedural rules. The ruling came in Tate Group Automotive, LLC v. Legacy Automotive Capital, LLC, Cause No. 25-BC11B-0020, after the Court completed an in camera review of ChatGPT “conversations” that the plaintiff had withheld from production. The decision continues the developing judicial dialogue on AI discoverability, siding with two federal courts that have found no waiver and expressly splitting from another federal court that reached the opposite conclusion. In doing so, the Court rejected the argument that a non-lawyer’s sharing of information with ChatGPT necessarily waived the attorney work product privilege.

The Court’s Ruling

Judge Dorfman agreed with two recent federal decisions cited by the plaintiff: Warner v. Gilbarco, Inc., 2026 WL 373043 (E.D. Mich. Feb. 10, 2026) and Morgan v. V2X, Inc., 2026 WL 864223 (March 30, 2026). Both of those courts recognized that work product protections are waived by disclosure to an adversary—or in circumstances that substantially increase the likelihood that an adversary will obtain the materials—and that sharing information with an AI tool does not meet that threshold.

Judge Dorfman also noted that the Texas Rules of Civil Procedure set a distinct (and arguably broader) standard than the federal rules. Under Texas Rule of Civil Procedure 192.5(a)(1), protectable work product includes “material prepared or mental impressions developed in anticipation of litigation or for trial by or for a party.” The Court found that this language “plainly appear[s] on [its] face to extend that protection to [the plaintiff’s] ChatGPT conversations.”

The Court’s holding is an express rejection of the contrary reasoning set forth in United States v. Heppner, 2026 WL 436479 (S.D.N.Y. Feb. 17, 2026), and highlights the jurisdiction by jurisdiction approach developing in this area.

The Protection Has Limits

The ruling was not a blank check. Consistent with the approach in Morgan, the Court ordered the plaintiff to disclose to the defendants all discovery materials or products that it had shared with ChatGPT, including any materials produced under the protective order. In other words, even if your AI prompts are shielded, the underlying materials you fed into the tool may not be. And, importantly, a party may violate a protective order by disclosing protected materials with an AI tool.

The Court also ordered production of certain pages from the in camera submission that did not constitute protectable work product. And it recommended that the parties confer and negotiate amendments to their protective order to make “unquestionably clear whether, how, and to what extent” confidential information may be shared with AI tools or other language model systems.

Big Picture

The case law in this area is still in its infancy. As Judge Dorfman acknowledged, the issue is a “novel one,” with all cited authorities dating from 2026 alone and at least one court describing its own ruling as “a question of first impression nationwide.”

The federal courts are also not aligned. Warner and Morgan say sharing with AI does not waive work product protection. Heppner says it does. And the Texas Business Court has weighed in on the pro-protection side of the ledger, bolstered by the broad language of the Texas rules.

The Takeaway

While the Texas Business Court’s decision may give some comfort to litigants who use AI, such litigants and their counsel still must proceed with caution. Courts are actively working through these questions in real time, and there is no settled consensus. Parties using AI tools in connection with litigation should assume their use may be scrutinized and should preserve records of what they share with AI platforms.