AI Court Orders and Ethics in Texas

Short Answer

Texas lawyers may use generative AI, but the permission comes wrapped in duties. The Northern District's Local Civil Rule 7.2(f) requires a first-page disclosure on any brief prepared with generative AI, and a filing without the disclosure certifies that no part of it was AI-prepared. Individual judges layer their own conditions on top: Judge Boyle bars AI-generated text in filings before her entirely, while Judge Starr has dropped his former separate AI certification now that the district rules cover disclosure. The Southern District cautions under Rule 11 without banning AI use, and an Eastern District standing order requires a certificate naming the tool and how it was used. The Texas Supreme Court has preliminarily approved rule amendments that would make every signer represent that the filing was reviewed and verified, with fabricated citations in view, but final adoption had not been located as of this writing. Ethics duties of competence, confidentiality, candor, and supervision apply to AI use under the disciplinary rules and Texas Professional Ethics Committee Opinion 705. On privilege, two February 2026 federal orders point in different directions because the facts differ: self-directed consumer AI use without counsel direction or confidentiality safeguards lost protection in United States v. Heppner, while a pro se litigant's ChatGPT litigation preparation stayed protected as work product in Warner v. Gilbarco.

Which Rules Apply

Texas procedure: proposed amendments to Tex. R. Civ. P. 13 and Tex. R. App. P. 9 and 52, Supreme Court of Texas Misc. Docket No. 26-9054 (preliminary approval June 26, 2026; not final).

Texas ethics: Texas Disciplinary Rules of Professional Conduct 1.01 (competence), 1.05 (confidentiality), 3.03 (candor toward the tribunal), 5.01 and 5.03 (supervision); Texas Professional Ethics Committee Opinion 705 (February 2025).

Federal: Fed. R. Civ. P. 11; N.D. Tex. Local Civil Rule 7.2(f) and Local Criminal Rule 47.2(e); S.D. Tex. General Order 2025-04; E.D. Tex. Judge Love standing order on generative AI (April 9, 2025); Fed. R. Civ. P. 26(b)(3) (work product).

The Northern District Rule

Local Civil Rule 7.2(f) of the Northern District of Texas provides that a brief prepared using generative artificial intelligence must disclose that fact on its first page under the heading "Use of Generative Artificial Intelligence." If the presiding judge directs, the filing party must disclose the specific parts prepared with generative AI. The rule defines generative artificial intelligence as a computer tool, by whatever name, capable of generating new content such as images and text in response to a submitted prompt by learning from a large reference database of examples. A party who files a brief without the required disclosure certifies that no part of the brief was prepared using generative AI. The current local rules took effect September 2, 2025. N.D. Tex. Local Civil Rules

Judge-Specific Requirements

Two judges illustrate how differently the district's judges implement the rule.

Judge Jane Boyle's requirements, effective September 1, 2026, prohibit any text generated by artificial intelligence in any portion of any pleading, motion, brief, or other filing made by an attorney before her. All language in attorney-signed filings must be drafted by a human attorney and checked for accuracy by a human attorney. Using AI to identify errors in documents is acceptable but must be disclosed under Local Rule 7.2(f), and any resulting edit must be made and verified by a human attorney. Every filing must include a signed certificate stating that the text was generated in its entirety by a human attorney; failure to include it may result in the filing being stricken. Pro se parties are equally responsible for compliance, and noncompliance may result in litigation sanctions or Rule 11 sanctions. Judge Boyle requirements

Judge Brantley Starr's requirements page states that he no longer requires a certification on the use of artificial intelligence, in light of the court's adoption of Local Civil Rule 7.2(f) and Local Criminal Rule 47.2(e), which require disclosure of generative AI use in documents lawyers prepared using generative AI. His former separate certification requirement should not be described as current. Judge Starr requirements

Other Texas Federal Courts

The Southern District of Texas, in General Order 2025-04 entered May 7, 2025, cautions attorneys and self-represented litigants against submitting pleadings, motions, or other papers drafted with generative AI without checking them for accuracy, and provides that any signer will be held responsible for the filing's contents under Rule 11 regardless of whether generative AI drafted any portion. It imposes no AI ban and no disclosure certificate. S.D. Tex. General Order 2025-04

In the Eastern District of Texas, Judge Love's standing order of April 9, 2025 requires every party, represented or pro se, to include with all filings a "Certificate of Generative Artificial Intelligence Usage" disclosing whether generative AI was used, the specific tool used, how the tool was used in preparing the document, and the party's certification that the accuracy of any AI-drafted or AI-assisted portion was checked, including factual background, citations, and legal authority. A filing without the certificate is presumed to certify that no part was AI-prepared. The order holds the filer responsible under Rule 11 and the rules of professional conduct and warns of possible sanctions. Judge Love standing order

The Pending Texas Statewide Amendments

On June 26, 2026, the Supreme Court of Texas gave preliminary approval to amendments to Tex. R. Civ. P. 13 and Tex. R. App. P. 9 and 52, with repeal of Tex. R. App. P. 57. The proposed Tex. R. Civ. P. 13(c)(1) and Tex. R. App. P. 9.1(d)(1) would make the signer represent that the signer reviewed the filing and verified its accuracy, including that legal citations, quotations, and holdings are not fabricated. The comments state these provisions were adopted principally to address concerns about the improper use of artificial intelligence in filings, which increases the risk of fabricated authorities appearing in court filings, and that verification matters regardless of whether AI was used. The order invited public comments through September 1, 2026 and expected the amendments to take effect October 1, 2026, but the court reserved the right to change them and to issue a finalizing order. A final adoption order had not been located as of October 11, 2026, so the proposal should not be treated as effective. The current consolidated Tex. R. Civ. P. 13 and Tex. R. App. P. 9 continue to govern. Misc. Docket No. 26-9054

Ethics Duties for AI Use

The disciplinary rules supply the enforceable duties. Rule 1.01 requires competent representation. Rule 1.05 protects client confidential information. Rule 3.03 requires candor toward the tribunal. Rules 5.01 and 5.03 impose supervision duties for lawyers and nonlawyer assistants, which extend to AI use by others in the firm. Texas Professional Ethics Committee Opinion 705, issued in February 2025 at the request of the State Bar's Taskforce on Responsible AI in the Law, applies these duties to generative AI. On competence, a lawyer who uses generative AI must have a reasonable and current understanding of the technology, including the risk of hallucinations and the limitations of the model's data. On confidentiality, the lawyer must be reasonably satisfied the program will not reveal client confidential information, considering terms of service, data security, and the risk that inputs are stored or used for training; where confidential information will be disclosed, client consultation and consent are required at a minimum. On verification, the lawyer remains responsible for accuracy and must verify AI output before using it in client work or filings. On fees, a lawyer billing hourly may not charge for time not actually spent, though reasonable AI costs may be passed through with the client's agreement. The opinion cites the Mata v. Avianca sanctions as the cautionary example. Opinion 705

Privilege and AI Tools: Heppner and Warner

Two federal orders issued in February 2026 show how privilege analysis turns on the facts of the AI use.

In United States v. Heppner, No. 25 Cr. 503 (JSR) (S.D.N.Y.), Judge Rakoff ruled from the bench on February 10, 2026 and issued a written opinion on February 17, 2026, holding that roughly thirty documents memorializing the criminal defendant's exchanges with the consumer AI platform Claude were protected by neither the attorney-client privilege nor the work product doctrine, and the government was permitted to inspect them. The defendant had used Claude on his own initiative after receiving a grand jury subpoena, without any direction from counsel, to prepare reports outlining defense strategy, and later shared them with counsel. The court held Claude is not an attorney, so no attorney-client relationship existed; the communications were not confidential because Anthropic's privacy policy put users on notice that inputs and outputs could be used for training and disclosed to third parties including governmental authorities; and later sharing the materials with counsel could not make non-privileged communications privileged. Work product failed because the materials were not prepared by or at the behest of counsel and did not reflect counsel's strategy. The court noted a counsel-directed arrangement might be analyzed differently but did not decide it. Heppner written opinion

In Warner v. Gilbarco, No. 2:24-cv-12333 (E.D. Mich.), Magistrate Judge Patti denied on February 10, 2026 the defendants' motion to compel "all documents and information concerning [the plaintiff's] use of third-party AI tools in connection with this lawsuit." The pro se employment-discrimination plaintiff had used ChatGPT to prepare litigation materials. The court held the materials protected as work product under Fed. R. Civ. P. 26(b)(3), which a pro se litigant may assert, and rejected the waiver argument: work-product waiver requires disclosure to an adversary or in a way likely to reach an adversary, and "ChatGPT (and other generative AI programs) are tools, not persons, even if they may have administrators somewhere in the background." The court also noted the defendants had no evidence the plaintiff uploaded protective-order materials to the platform. Warner order, ECF No. 94

The contrast is instructive rather than contradictory. Heppner turned on the absence of counsel direction and the absence of confidentiality safeguards under the platform's terms. Warner turned on work product's higher waiver threshold and the absence of any disclosure to an adversary. Neither order decides the counsel-directed enterprise-AI case, where written confidentiality, no-training, and access commitments are in place.

What Is Unsettled

Whether the proposed Tex. R. Civ. P. 13 and Tex. R. App. P. 9 and 52 amendments take effect as preliminarily approved and on what date. Whether enterprise AI terms preserve confidentiality for privilege purposes, whether a Kovel-type arrangement can extend privilege to AI-assisted analysis, and whether Heppner's reasoning will carry into civil cases and Texas courts. How judge-specific AI orders will interact with the statewide verification representation if the amendments are adopted.

A Verification Routine

Pull every cited authority from an official or reliable database and read the cited passage. Check every quotation word for word. Confirm subsequent history. Keep a record of verification for filings in courts with certification requirements. Do not enter confidential client information into tools without appropriate confidentiality terms. For every filing, check the district's local rules and the assigned judge's requirements page on the day of filing, because judge-specific AI conditions change.

Practice Checklist

Before filing or advising on AI use in Texas litigation:

• Determine which court the filing is in and read that court's current AI rule or order: N.D. Tex. LR 7.2(f) disclosure, S.D. Tex. General Order 2025-04 Rule 11 caution, E.D. Tex. Judge Love's certificate, or the applicable local rule elsewhere.
• Read the assigned judge's requirements page for AI conditions on top of the district rule. Before Judge Boyle, no AI-generated text may appear in the filing and a signed human-draft certificate is required. Before Judge Starr, no separate AI certification is currently required.
• If any part of a brief was prepared with generative AI in the Northern District, place the "Use of Generative Artificial Intelligence" disclosure on the first page. Omitting it certifies no AI was used.
• Verify every citation, quotation, and holding against the actual authority before signing. Treat the proposed Tex. R. Civ. P. 13 and Tex. R. App. P. 9 verification representation as the direction of travel even while the amendments remain unadopted.
• Apply Opinion 705 before putting client information into an AI tool: understand the tool's data practices, confirm confidentiality protections in writing where possible, and obtain client consent where the rules require it.
• Supervise AI use by others in the firm under Rules 5.01 and 5.03, and do not bill hourly clients for time not actually spent.
• For privilege, do not assume consumer AI use is protected. Heppner lost protection where use was self-directed and the platform's terms permitted data use and disclosure. Warner kept work-product protection where use was litigation preparation and nothing reached an adversary. Document counsel direction, limit platform data rights by contract, and keep AI-assisted litigation preparation out of adversaries' hands.
• Watch the Texas Supreme Court's docket for a final adoption order on the Rule 13, 9, and 52 amendments before treating the verification representation as effective.