Is It Appropriate to Use AI in an Article About the Misuse of AI?

As courts continue to receive filings prepared with the assistance of artificial intelligence (“AI”) from both pro se litigants and attorneys, there is an increasing burden to scrutinize filings, including allegations, motions, and purported evidence, and to verify the accuracy and reliability of the submissions. If a litigator or court identifies inaccuracies, what could happen next? Below are several 2026 federal court decisions addressing the misuse of AI tools and the resulting sanctions.

1. Booker v. Kroger Co., N.D. Ga., Atlanta Division, Case No. 1:26-cv-02006, Opinion Filed Aug. 28, 2026; Unreported Opinion

After issuing an order to show cause and conducting a hearing, the court found that plaintiff’s counsel violated Federal Rule of Civil Procedure 11 (Rule 11) by:

1. Submitting nonexistent citations and misrepresenting legal propositions in court filings;
2. Falsely stating that no AI tools had been used in preparing the briefs, including representing on the record that “[he] did not use artificial intelligence”; and
3. Misleading the court by feigning a lack of understanding regarding the court’s inquiry into AI use.

(Op. at 3-4).

The court identified four hallucinated citations in counsel’s filings and imposed sanctions of $2,000 per citation, totaling $8,000. (Op. at 4). The court further indicated that it could impose additional sanctions unless counsel provided satisfactory proof of completing continuing legal education (CLE) or other training concerning the ethical use of AI. (Op. at 5).

In reaching its decision, the court quoted from its Standing Order regarding artificial intelligence:

Many courts and individual judges around the country have entered specific rules or procedures concerning the use of artificial intelligence tools. My rule is simple: You can use whatever AI tools you like, but only human beings will be held responsible for the outcome. So, if a pleading or paper that is filed in a case contains factual or legal errors, the party and lawyers who filed it will be held responsible. This includes each and every lawyer listed on the signature block of the filing. Govern yourself accordingly, as they say. See Order: SDG_CVStandingOrder_20260210

The court also observed that, although counsel ostensibly accepted responsibility, he simultaneously attempted to shift blame by referring to “this office” as the drafter of the filings and attempting a “disingenuous” explanation for lying to the court. (Op. at 3). The case remains pending.

2. Gregoire v. Bd. of Trs. of San Francisco Bay Area Rapid Transit Dist., N.D. CA, Case No. 25-cv-02150, 2026 U.S. Dist. LEXIS 95767, 2026 WL 1140876; Opinion

The Northern District of California issued an order to show cause after discovering that plaintiff’s counsel submitted an opposition brief citing three apparently nonexistent cases. The court expressed concern that the inaccuracies suggested “the possible use of generative artificial intelligence tools resulting in hallucinations.” (Op. at *1).

The court emphasized that, under Federal Rule of Civil Procedure 11(b), counsel of record remains responsible for ensuring the accuracy and reliability of all filings, including content generated with the assistance of AI tools. Citing Oneto v. Watson, 808 F. Supp. 3d 974 (N.D. Cal. 2025), the court reiterated that parties have a duty to independently verify the accuracy of AI-generated content before submitting it to the court. (Op. at *2).

The court provided counsel with an opportunity to respond by ordering counsel to file a declaration identifying each challenged statement and citation, explaining the source of each citation, disclosing whether any generative AI tool was used, describing the steps taken to verify accuracy, and supplying corrected citations where appropriate. (Op. at *2-3).

A subsequent review of the court’s docket reflects that sanctions were imposed on May 6, 2026, including:

• A monetary sanction of $1,000;
• Service of the sanctions order upon counsel’s client; and
• Certification of attendance at continuing legal education addressing the ethical use of artificial intelligence.

Additional orders entered on June 1, 2026, indicate that counsel failed to certify full compliance with the court’s Order to Show Cause. As a result, the court imposed a sanction of $50 per day until full compliance was certified and granted the defendant’s motion for attorney fees in the amount of $2,362.50. The case remains pending.

3. Brownfield v. Cherokee County School District No. 35, E.D. Oklahoma, Case No. CIV-21-312, Opinion Filed March 19, 2026, 2026 U.S. Dist. LEXIS 57452; 2026 WL 777230; Opinion

The Eastern District of Oklahoma imposed monetary sanctions on a pro se plaintiff who relied upon AI-generated fictitious citations in a Rule 11 sanctions motion filed against the defendants.

The court found that the motion contained eight citations that either:

• Did not exist;
• Did not support the legal proposition for which they were cited;
• Did not contain the quoted language attributed to them;
• Reached conclusions contrary to those represented; or
• Involved facts inconsistent with the plaintiff’s descriptions.

(Op. at *2).

The plaintiff acknowledged using a generative AI tool to assist with legal research and further admitted that he failed to verify the accuracy of the resulting citations. (Op. at *3).

The court applied the two-step Rule 11 analysis set forth in Adamson v. Bowen, 855 F.2d 668 (10th Cir. 1988). First, it determined that the plaintiff violated Rule 11(b) by failing to conduct a reasonable inquiry into the legal authority cited in his motion. Second, the court evaluated an appropriate sanction and awarded $500 in attorney fees incurred in responding to the motion. (Op. at *5).

The court found two mitigating factors: the pro se litigant had proactively completed a two-month AI training program costing approximately $1,800, and he represented that he would not rely on generative AI tools in future proceedings before the court. (Op. at *9).

Subsequent docket review indicates that the matter proceeded to trial and is currently on appeal.

Emerging Standards and Expectations 

Collectively, these decisions reflect several principles governing the use of artificial intelligence in litigation:

1. Federal Rule of Civil Procedure 11(b) requires attorneys and unrepresented parties to conduct a reasonable inquiry under the circumstances before submitting representations to the court, including representations prepared with the use of artificial intelligence.
2. Deliberate misrepresentations regarding AI use or attempts to deflect accountability for AI-generated content may result in harsher sanctions than negligent failures to verify the accuracy of that content.
3. Courts may respond to the misuse of AI-generated content by imposing monetary sanctions and other remedial or educational measures, such as requiring ethics or CLE training.
4. Although remedial measures may mitigate the severity of sanctions, they may not eliminate them, particularly when taken only after the court has discovered the misuse.

Courts nationwide are increasingly adopting standing orders, local rules, and guidance addressing the use of artificial intelligence in litigation. Many of these developments are tracked in the publicly available Artificial Intelligence Court Order Tracker. Given the growing reliance on artificial intelligence in the practice of law, sanctions for its misuse are likely to become more common and more severe, particularly when counsel or pro se litigants appear to misrepresent their reliance on unverified AI-generated content.

So, Is It Appropriate to Use AI for an Article About the Misuse of AI? 

The answer is yes, provided the technology is used responsibly.

Some AI assistance was used in preparing this article, but all substantive content was independently reviewed and verified by the author. Specifically:

1. LexisNexis Protégé™, a closed, enterprise-grade generative AI platform, was used as an initial research tool to identify potentially relevant cases. Five decisions issued in 2026 were identified, and all were independently reviewed by the author. Two cases were ultimately determined to be outside the scope of this article.
2. Case summaries generated by the platform were reviewed but largely disregarded. The summaries presented in this article were independently drafted by the author.
3. Court docket activity and procedural developments were independently reviewed and verified by the author before inclusion.
4. A closed-environment Microsoft Copilot system was used to assist with final editing and proofreading. Although some of its suggestions were incorporated, others were not, and reviewing and removing the rejected suggestions may have taken longer than editing the article without AI assistance.

The critical distinction is not whether AI was used, but whether the resulting work was subjected to meaningful human review, verification, and professional judgment before publication.

Jenifer E. Novak and Frantz Ward’s Litigation Practice Group will continue to monitor developments and recommendations for using artificial intelligence in litigation. Should you have any questions, please reach out to our team.