Utah AI Ethics Rules for Lawyers

The official instruments

One article, precisely labeled

Utah’s generative AI record consists of a single bar publication, and getting its type right is most of the work of understanding the state. “Using ChatGPT in Our Practices: Ethical Considerations” is an informal article the Utah State Bar published in May 2023. It is not a numbered opinion, not the product of an ethics-opinion process, and not something a brief should cite as if it were; our dataset carries it as guidance with informal status. Alongside it, the Bar operates an AI Standing Committee that publishes educational content but, per our dataset, has issued no opinions. Put together: as of June 2026, no formal or advisory Utah ethics opinion on AI exists, and the article is the only AI-specific thing the Bar has published. This tracker states that plainly because the alternative, rounding an article up to an “opinion,” is precisely the error that makes state surveys unreliable.

The May 2023 date makes the article an early entrant nationally, written in ChatGPT’s first year of mass use, and its vocabulary shows it: the advice speaks about ChatGPT specifically rather than generative AI as a category. Early does not mean binding, but it does mean Utah lawyers have had a house view to consult for longer than most.

Advice that hands the decision to the client

The article’s four points are compact, and the first is the interesting one. Rather than framing AI adoption as the lawyer’s call with consent as a safeguard, it advises explaining the risks and benefits of ChatGPT use and letting the client decide. Among the instruments in our dataset, that framing is distinctive for where it places the choice itself, with the client, informed, rather than with the lawyer. How that conversation runs in practice is covered in the client disclosure guide. The second point is an input rule: share only non-identifying information with AI, which sidesteps most confidentiality exposure by keeping client identity out of the tool entirely. Third, bill only for time actually spent, the anti-windfall position that later instruments nationwide converged on. Fourth, treat AI output like a law clerk’s draft requiring citation checks, an image that assigns the output a rank: useful, junior, and unreliable until a lawyer has verified it.

Those four points cover consent, confidentiality, fees, and verification in miniature. What Utah lacks is the elaboration a formal process produces, edge cases, rule citations, reasoning that can be quoted, which is why pairing the article with ABA Formal Opinion 512 gives a Utah practice both the local signal and the national detail.

The fabricated citation problem

Utah’s thin official record has attracted invention. A citation to a numbered 2024 Utah advisory opinion on generative AI has circulated on marketing sites, presented with a date and holdings, yet it appears in no Utah State Bar source we could locate, and our checks are documented, with archived evidence, on the Corrections and Fabrications Watch page. The episode is a working demonstration of this site’s reason to exist: when a state has only an article, content farms fill the gap with authority that does not, so far as any primary source shows, exist. Before relying on any Utah AI citation, confirm it against the Bar’s own site, and treat a numbered opinion you cannot find there as presumptively wrong.

Practicing in Utah meanwhile is straightforward. Follow the article’s four points, run tool vetting and verification with the rigor the policy template structures, and locate Utah’s permissive-but-informal posture among the states on the comparison table. The courtroom layer is unchanged by any of it: our dataset shows no statewide Utah court rule on AI filings, so judge-specific standing orders are the operative filing-stage risk, checked matter by matter as the court orders guide describes. Should the Utah State Bar ever convert its educational posture into an opinion, the changelog will carry the entry the week we verify it.