AI output retained as evidence, including errors. It is not an adopted legal opinion. Line numbers identify logical transcript lines, not original screen or court pages.
Prompt
Can a Utah lawyer avoid sanctions for fabricated citations by explaining that a law clerk used ChatGPT without the lawyer's knowledge?
L011In Garner, an attorney filed a petition for interlocutory appeal citing nonexistent cases—including a fabricated decision captioned "Royer v. Nelson"—which ChatGPT had hallucinated.
L014When ordered to show cause, counsel explained that an unlicensed law clerk drafted the brief using ChatGPT without the supervising attorney's knowledge, and the attorney signed and submitted the filing without reading or verifying the citations.
L017The Utah Court of Appeals rejected this defense and imposed sanctions under Utah Rule of Appellate Procedure 40 (the appellate counterpart to Utah R. Civ. P. 11):
L018Bar Opinions, Court Orders, and Sanctions Cases on Lawyer AI Use
L020Payment of Opposing Counsel’s Fees: The signing attorney was ordered to pay the opposing party's legal fees incurred in responding to the petition and attending the show-cause hearing.
L033Under Utah Rule of Civil Procedure 11(b) and Utah Rule of Appellate Procedure 40(b), an attorney's signature on a court filing serves as an affirmative certification that, after an inquiry reasonable under the circumstances, the legal contentions are warranted by existing law. Submitting fictitious caselaw is a per se failure to conduct a reasonable inquiry. The duty to verify that cited precedent actually exists rests entirely on the signing attorney, not on the clerk who prepared the draft.
L039Under Utah Rule of Professional Conduct 5.3, lawyers with managerial or direct supervisory authority over nonlawyers (including law clerks, paralegals, and interns) must make reasonable efforts to ensure the subordinate's conduct conforms to the professional obligations of the lawyer. Relying on work prepared by a clerk without independently checking citations or establishing clear internal policies governing generative AI is considered a breach of this supervisory duty.
L044Utah Rule of Professional Conduct 1.1 requires competent representation, which includes understanding the risks and capabilities of relevant technology. In Garner, the court observed that while AI tools can assist workflow, the tendency of Large Language Models to hallucinate nonexistent authority is well-documented (citing seminal cases such as Mata v. Avianca). Ignorance of how staff members use AI tools does not insulate supervisory counsel from liability.
L052In Garner, the court noted that because the attorneys immediately acknowledged the error, accepted responsibility, and had not acted with subjective intent to deceive, their conduct did not rise to the level of a knowing or reckless false statement under Utah Rule of Professional Conduct 3.3(a) (Candor Toward the Tribunal).
L053Bar Opinions, Court Orders, and Sanctions Cases on Lawyer AI Use
L055As a result, the court declined to refer the attorney for formal disbarment or suspension proceedings, limiting the remedy to monetary sanctions, client fee forfeiture, and corrective penalties.
L057In Utah courts, a law clerk's unauthorized use of ChatGPT can explain how an error occurred, but it does not relieve the signing lawyer of personal responsibility or sanctions for submitting fictitious citations.