On September 8, 2026, the Missouri Court of Appeals, Eastern District, issued an opinion in State ex rel. Schnell v. Richardson where the Court discussed the landscape for using AI and the repercussions that can arise from misusing it in Missouri Courts. The case arose initially from a family law dispute seeking termination of parental rights and the adoption of a minor.
There were cases pending in multiple Counties involving the family. The attorney for one of the parties suggested in the briefing that Missouri Law required that all the cases be consolidated into the same County and Circuit and be heard by the same judge and moved for transfer of his case. In his briefing filed with the court the attorney argued for transfer of the case to another venue. In support of his argument, the attorney relied on two cases which he used as support for the claim that the pending cases must be heard in the same jurisdiction because they involved the same family.
The Court looked at the briefing and attempted to locate the cases replied upon by the attorney. The Court found cases corresponding in name to the cases offered by the attorney – but the named cases did not have citations which matched. The Court looked further and discovered that the citations given did not correspond to the named cases. It appears the case names given were associated with actual cases, but the citations to those named cases pointed to different cases. Further, the cases to which the citations pointed were not on topic for the legal issues presented or, more notably, as represented by the attorney.
Given this discrepancy, the Court issued a show cause order directing the attorney to explain the discrepancies. The attorney explained that they had used AI to research the issue and had failed to verify the accuracy of the citations before using them. The Court noted that the attorney was “earnest and apologetic” when the mistake came to light – but nevertheless pointed out that the citations misrepresented the law and falsely suggested that the attorney’s argument was supported by Missouri law. Further, what the Court described as “worse still”, was that the Court had acted on the misrepresented law and had asked the other party to respond to the attorney’s inaccurate briefing.
The Court then turned to Rule 55.03(b) which mandates that by making an argument to the Court an attorney is “certifying” that after reasonable inquiry the legal contentions made are “warranted by existing law…” The Court found that (1) because the citations offered by the attorney were inaccurate that the attorney’s arguments were not warranted by existing law and (2) because the attorney did not check the citations for accuracy, they could not have made the reasonable inquiry required by the rule.
The Court then noted the well documented tendency of AI to fabricate case law and citations – i.e. create ‘hallucinations’, and additionally tendency of AI to simply cites cases which do not support the proposition offered or misstate holdings entirely. In this context the Court patently held that:
Given the well-known tendency of generative AI models to make fabrications, it is objectively unreasonable for any attorney or self-represented party to fail to verify AI’s accuracy. For this reason alone, sanctions are appropriate
The Court gave no leeway and offered no exceptions or potential justifications for an attorney or self-represented party failing to check the accuracy and veracity of AI generated citations. The Court’s rule is simple: failing to check AI citations is a violation of Rule 55.03(b).
Likewise, the Court cautioned that improper or careless use of AI could also violate an attorney’s ethical obligations under Rule 4-1.1 and 4-3.1, which require competent representation and prohibit attorneys from bringing claims without a basis in law and fact. The Court also pointed out the risk of ethical violations under Rules 4-5.4 and 4-8.4, which require attorneys to maintain independent professional judgment and prohibit dishonesty or misrepresentations prejudicial to the administration of justice.
However, in notable commentary on the use of AI, the Court did not condemn AI use and, in fact, stated that there is “nothing inherently problematic” about using AI. It described the use of AI as commonplace and “profoundly beneficial” which “when used properly” can be an effective tool for an attorney
The Court sanctioned the attorney, ordering the attorney to pay monetary damages to the other party for costs and attorney’s fees incurred. The holding makes clear the risks associated with the use of AI research and generative tools – most notably, the failure to thoroughly check the basis for the output of any AI model.
The Court’s holding in Schnell is the most recent in what is a growing body of cases and commentary addressing AI use in Missouri. Other examples include the Missouri Eastern District court to issue a monetary sanction where twenty-two of twenty-four citations in an appellate brief were fictitious, calling the filing “a flagrant violation of the duties of candor” owed to the court. Kruse v. Karlen, 692 S.W.3d 43 (Mo. App. 2024). In Mills v. City of St. Louis, 2026 WL 251781 (E.D. Mo. Jan. 30, 2026), the court sanctioned a self-represented plaintiff under Federal Rule 11 and dismissed the action with prejudice, noting that “[e]very filing in a federal court that contains citations to phony case law amounts to a violation of Rule 11(b).” Missouri Circuits, such as the Seventh Judicial Circuit in Clay County, are beginning to enact local rules that require parties to disclose generative AI use.
Without question attorneys should check their citations and verify the accuracy and veracity of the authority cited whenever utilizing AI tools. When responding to any motions and supporting memorandum and/or briefing counsel should carefully review statements of law with case citations for accuracy and if blatantly off the mark file the appropriate motion for a remedy and if called for sanctions. The cost of responding to unsupported authority and false statements of the law based upon simply reciting a printout from an AI source falls on the responding party long before any Court may intervene.