Why It Matters
An Arizona appeals court has ruled that attorneys and self-represented litigants who cite non-existent cases generated by artificial intelligence face financial penalties, regardless of intent. The decision marks the state judiciary’s first published guidance on AI accountability in legal work and establishes a clear standard for professional responsibility in a growing area of courtroom practice.
What Happened
The Arizona Court of Appeals issued a unanimous opinion addressing a self-represented litigant’s use of generative AI in preparing a legal brief. Jeffrey Kotchka appealed a probate decision involving his sister Kim Dineen, who sought to administer their mother’s estate after Gloria Acciavatti died in June 2023 at age 93. Dineen filed her petition in February 2024 and was appointed personal representative by a Maricopa County judge in July 2025.
In his opening brief, Kotchka included eight citations to support his appeal. The three-judge panel, led by Judge Brian Furuya, found six of those citations deficient—two of which cited cases that do not exist at all. Kotchka acknowledged relying on generative AI to draft his brief and admitted he had not verified the citations using publicly available legal research tools or sought professional assistance.
The court ruled that submitting hallucinated case citations constitutes sanctionable conduct whether intentional or accidental. Kotchka was ordered to pay a portion of Dineen’s attorney’s fees that could be traced directly to the false citations, and his appeal was rejected.
By the Numbers
Eight — citations included in Kotchka’s opening brief
Six — deficient citations identified by the court
Two — non-existent cases cited
July 2025 — month Dineen was appointed personal representative of the estate
Zoom Out
Generative AI hallucinations—instances where the technology produces confident but false information—have emerged as a significant issue across professional fields. In law, the problem has gained particular attention after several high-profile cases in which attorneys cited fabricated court opinions generated by ChatGPT and similar tools. Federal courts and state bar associations nationwide have begun grappling with how to regulate AI use in legal practice, with some jurisdictions imposing disclosure requirements or outright restrictions.
Arizona’s decision provides one of the earliest state appellate court rulings establishing clear consequences for AI-generated fiction in court filings. The ruling is notable for holding that negligence—failing to verify citations—carries the same penalty as intentional misconduct, placing the burden on users to validate AI output before submission.
What’s Next
The decision is likely to influence how Arizona courts and the state bar association approach AI disclosure rules and standards. Other states facing similar cases may look to Arizona’s framework as a model for sanctioning conduct. Litigants using generative AI for legal research or brief writing now face explicit notice that independent verification of citations is a mandatory professional responsibility, not an optional safeguard. The ruling may also prompt broader conversations about whether attorneys require specific training or certification before deploying AI tools in client representation.