Published opinion, 15 July 2026 · source · PDF · 10 pages
Because his arguments are unsupported by any valid authority, we affirm. ¶2 More concerning, however, is Mr. Kotchka’s failure to ensure the validity of legal citations in his opening brief that he obtained through use of generative artificial intelligence (“Gen-AI”), including his inclusion of hallucinated authorities. Because of these misrepresentations of law, we impose sanctions under Arizona Rule of Civil Appellate Procedure (“ARCAP”) 25, Arizona Revised Statutes (“A.R.S.”) Section 12-349, and the court’s inherent powers to manage its affairs.
KOTCHKA Opinion of the Court 4 include the use of cases for principles that are not supported by the case, mis-citation to wrong pages or paragraphs, misstatement of rules enunciated in cases, and misrepresentation of case facts. Most concerning, the brief cites two cases that simply do not exist. ¶9 Because of the brief’s inaccurate and false citations, we strike each such citation. Significantly, the brief contains no other supportive legal authority for its arguments. ¶10 In his written explanation in response to our order to show cause, Kotchka stated he relied on Gen-AI in writing his brief and blames it for the inaccuracies and falsities of his citations.
II. The Use of Hallucinated Case Citations Constitutes Abuse of Gen- AI and Is Sanctionable Conduct. ¶11 Because the issue of Gen-AI inaccuracies and hallucinated authorities in court filings —whether included in those filed by self- represented litigants or attorneys —poses a recurring and increasing 1 These materials are accessible, without cost, through the court’s official website: coa1.azcourts.gov, by clicking on the links under the menu labeled “Self -Represented.” They include the court’s “Guides for Self - Represented Litigants,” which advise of the need to read and follow applicable rules. DINEEN/SHIBATA v.
Gen-AI sometimes—perhaps even frequently —produces inaccurate or meritless legal theories and research results. Included among those inaccuracies are citations to legal authorities that simply do not exist, a phenomenon commonly known as “hallucination.” Thus, careless use of Gen-AI can lead to creation of briefs and other filings which present meritless legal theories or cite misleading, fake, or nonexistent legal authorities, an issue that is rapidly becoming common in our courts. An ever-growing body of evidence and experience demonstrates that Gen-AI is woefully inadequate at reliably providing accurate and trustworthy citations to valid legal authorities.
Cl. 163, 169 (2025) (“It is no secret that generative AI programs are known to ‘hallucinate’ nonexistent cases, and with the advent of AI, courts have seen a rash of cases in which both counsel and pro se litigants have cited DINEEN/SHIBATA v. KOTCHKA Opinion of the Court 6 such fake, hallucinated cases in their briefs.”); Hardy v.
See also D. Charlotin, AI Hallucination Cases https://www.damiencharlotin.com/hallucinations/ (last accessed July 13, 2026) (cataloguing 1,751 instances of AI-generated hallucinated content in legal documents described in court decisions from across the globe, 1208 of which were found in U.S. court filings). ¶14 Nor does it appear that Gen-AI tools will soon be able to fix the problem. As one federal circuit court put it, “[t]he hallucination problem has no end in sight, as AI ’s tendency to fabricate results arises from the training and structures of AI programs.” Fletcher v.
3d 227, 236 (App. 2026) (recognizing the “increasing incidence of hallucinated case citations generated by AI tools”). A major problem in using Gen-AI for legal work of this kind, however, is that many litigants do little to nothing to verify the accuracy of Gen -AI- produced legal work product, essentially abdicating their own duty to ensure accuracy in their filings, see Gomez, 231 Ariz.
Failure to act with reasonabl e diligence to check—and correct —any work product created using Gen -AI that is submitted to the courts constitutes abuse of Gen-AI tools in litigation. ¶17 We acknowledge that self-represented litigants in particular face challenges when using Gen -AI tools for legal research because such DINEEN/SHIBATA v. KOTCHKA Opinion of the Court 7 litigants may be unaware that these tools frequently generate hallucinations. Indeed, use of Gen -AI can give self-represented litigants a false sense of confidence in their filings.