# Utah Lawyer, Fabricated Citations, and the "My Clerk Used ChatGPT" Explanation ## Short answer The explanation is unlikely to **avoid** sanctions, but it can meaningfully **reduce** them. Utah's Rules of Professional Conduct place the ultimate responsibility on the signing lawyer, and the "I didn't know" framing collides with several supervisory and competence duties. That said, showing the fabrication was an AI hallucination generated by a subordinate—rather than deliberate fabrication by the lawyer—shifts the analysis from *intentional misconduct* toward *negligent process failure*, which courts treat more leniently. --- ## The rules that make the defense difficult ### Rule 5.3 – Duties Regarding Nonlawyer Assistance This is the central rule. A lawyer is responsible for a nonlawyer's conduct if the lawyer: - **(a)(1)** failed to make *reasonable efforts to prevent and remediate* conduct that would violate the Rules if done by the lawyer; or - **(b)** directed or, *with knowledge of the specific conduct*, ratified it; or - **(c)** as a partner or supervisory lawyer, directed or *had knowledge of* the conduct. Even granting that the lawyer had no actual knowledge of the specific hallucinated citations, **5.3(a)(1)** is an independent, *preventive* duty. If the firm had no AI-usage policy, no citation-verification step, and no review protocol for clerk-drafted filings, a tribunal can find the lawyer failed the "reasonable efforts to prevent" prong. The absence of a guardrail is itself a sanctionable act; the clerk's use of ChatGPT is the symptom, not the whole violation. ### Rule 1.1 – Competence (including technology) Utah's Rule 1.1 requires "knowledge of the law and its practice" and competence that includes "the technological skills" necessary to represent a client. A supervising lawyer who allows a clerk to run a brief through a generative AI tool with zero verification workflow will be hard-pressed to argue the firm met its competence obligation. ### Rule 1.3 – Diligence "Reasonably diligent" includes reviewing what is filed under your name. A blanket "I didn't read it" or "I didn't know the clerk ran it through GPT" is a weak diligence argument. ### Rule 3.3(a) – Candor Toward the Tribunal Submitting citations that do not exist is a direct 3.3 violation. The rule does not contain a "good-faith clerical error" exception. The lawyer's personal intent matters for *severity*, but the *act*—a fabricated citation in a filing under the lawyer's name—stands. ### Rule 8.4 – Misconduct 8.4(c) (disparagement, conduct involving dishonesty) and 8.4(d) (conduct prejudicial to the administration of justice) can both be implicated, and neither requires the lawyer's personal fabrication. --- ## Where the explanation *does* help Courts in the 10th Circuit, the District of Utah, and other jurisdictions that have recently confronted AI-hallucination briefs have drawn a line between: | Scenario | Typical treatment | |---|---| | Lawyer or firm *deliberately* fabricated or knowingly submitted fake cites | Harsh sanctions: suspension, fines, admission conditions, referral to ethics committee | | AI-generated phantom cites from a clerk's draft, no verification process | Sanctions for process failure: refile with verified cites, court costs, small fine, mandatory CLE on AI, written warning | | Same, but the firm *did* have a review policy and the clerk circumvented it | Reduced or no monetary sanction; focus on remediation | The "clerk used ChatGPT" narrative is most useful for moving the case from row 1 toward row 3. It demonstrates: - **No intent to deceive the court.** The lawyer can credibly argue there was no *purposeful* fabrication, only a negligent review gap. - **Prompt remediation.** If the lawyer self-reported, moved to withdraw the cites, and offered to bear costs, that weighs heavily in mitigation. - **Root cause is a system gap**, not a character flaw, which invites a corrective (policy adoption, CLE, supervised re-filing) rather than a punitive response. --- ## What Utah tribunals will likely probe 1. **Did the firm have any written AI-usage or citation-verification policy?** Absence is a significant negative. 2. **Did the lawyer review the brief before filing?** Even a cursory "I skimmed it" helps more than "I handed it off." 3. **Was the error self-discovered or caught by opposing counsel / the court?** Self-discovery and voluntary correction are strong mitigators. 4. **How many phantom citations?** One or two looks like a process miss; twenty-plus starts to look like a wholesale AI-generated draft with no human checking. 5. **Repeat pattern?** A single incident is treated very differently from a second one. 6. **Disclosure to the court.** If Utah's local rules or the specific judge's standing order require disclosure of AI assistance, failure to disclose is an *additional* 3.3 / 8.4 problem layered on top. --- ## Practical bottom line A Utah lawyer who walks in and says *"I didn't know a clerk ran the brief through ChatGPT"* will almost certainly still face **some** sanction—typically a refile order, an award of opposing counsel's costs, a fine in the low hundreds to a few thousand dollars, and a requirement to complete AI-ethics CLE. The explanation is a **mitigating narrative**, not an **exculpatory one**, because Rules 5.3(a)(1) and 1.1 impose duties that exist *independently of actual knowledge*. The lawyer's best posture is: - **Concede the filing error immediately and in writing.** - **Identify the exact AI tool, who used it, and when**, to demonstrate transparency. - **Present the firm's remedial steps** (new AI policy, citation-checking protocol, training, perhaps a supervising-attorney review gate for all clerk-drafted work). - **Request that the tribunal calibrate the sanction to the negligent-process tier** rather than the intentional-misconduct tier. - **Avoid over-arguing "it wasn't my fault."** Tribunals respond poorly to a lawyer who makes the clerk the sole villain while implicitly arguing the lawyer owes no duty. In short: the explanation changes *how much* the lawyer is sanctioned and *what the sanctions look like* (corrective vs. punitive), but under Utah's current rules it does not get the lawyer off the hook entirely. --- *This is legal information and analysis, not legal advice. Specific outcomes depend on the tribunal, the judge, the number and severity of the phantom citations, and the full factual record. A Utah-licensed attorney (ideally one who handles professional-responsibility matters) should review the specific situation before the lawyer responds to any court order or ethics inquiry.*