You said:
Can a Utah lawyer avoid sanctions for fabricated citations by explaining that a law clerk used ChatGPT without the lawyer's knowledge?
ChatGPT said:

Generally, no. A Utah attorney cannot automatically avoid sanctions for fabricated legal citations by claiming that a law clerk used ChatGPT without the attorney's knowledge.

The distinction is between responsibility for the filing and culpability for the misconduct.

Under Utah law, several rules matter:

1. Utah Rule of Civil Procedure 11

Under Utah Rule of Civil Procedure 11, an attorney's signature on a court filing certifies that, after an inquiry reasonable under the circumstances:

The legal arguments are warranted by existing law or a nonfrivolous argument for changing it.

The factual contentions have evidentiary support.

The filing is not presented for an improper purpose.

The attorney's certification is independent of who drafted the document.

If a clerk supplies fabricated citations and the attorney signs and submits the document without checking them, the question becomes whether the attorney conducted a reasonable inquiry.

Simply saying, "My clerk used ChatGPT and I didn't know," does not establish that the inquiry was reasonable.

2. Utah Rules of Professional Conduct 5.3 and 5.1

Rule 5.3 governs attorneys' responsibilities regarding nonlawyer assistants.

Lawyers with supervisory authority must make reasonable efforts to ensure that nonlawyer assistants' conduct is compatible with the lawyer's professional obligations.

The attorney may therefore face responsibility for inadequate supervision even without personally directing the clerk to use AI.

Rule 5.1 imposes related supervisory obligations concerning subordinate lawyers, which may apply if the clerk is a licensed attorney.

3. Utah Rules of Professional Conduct 3.3 and 1.1

Rule 3.3 requires candor toward the tribunal, including prohibiting knowing false statements of law and requiring correction in specified circumstances.

Rule 1.1 requires competent representation, including adequate preparation.

These rules matter differently:

Rule 3.3: An attorney who genuinely did not know the citations were fabricated may not have knowingly misled the court when filing. Once the problem is discovered, however, the attorney's duty to correct it becomes critical.

Rule 1.1: Failure to verify legal authorities can constitute inadequate preparation even without deliberate deception.

4. Can the clerk's unauthorized conduct mitigate sanctions?

Yes. This is where the defense could have some force.

A court assessing sanctions might consider whether:

The clerk deliberately concealed the use of ChatGPT.

The attorney maintained reasonable verification and supervision procedures.

The fabricated citations were difficult to detect despite reasonable diligence.

The attorney promptly corrected the filing and notified the court.

The attorney cooperated with the court's investigation.

These facts can affect whether sanctions are justified and their severity. They do not automatically eliminate responsibility.

There is also an important procedural distinction: Utah Rule 11 generally includes a 21-day safe-harbor procedure for party-initiated sanctions motions, while court-initiated sanctions follow a different procedure.

5. The federal analogy

In Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023), attorneys were sanctioned after submitting nonexistent cases generated by ChatGPT.

That case involved attorneys' own use of AI rather than an unsuspecting supervisor and a rogue clerk, so it is not directly controlling. Nevertheless, it illustrates that courts focus heavily on verification, representations to the court, and what attorneys do after errors are discovered.

Bottom line

An attorney can potentially avoid sanctions if the court finds that the attorney's inquiry and supervision were objectively reasonable despite the clerk's misconduct. But the attorney cannot avoid sanctions merely by proving ignorance of the clerk's AI use.

A fabricated citation is not necessarily proof of intentional dishonesty. It can, however, be evidence of a failure to conduct the reasonable inquiry required by Rule 11.

One qualification: I cannot verify Utah decisions issued through November 2026 because live web search is unavailable in this conversation. The rules above establish the general framework, but I would not characterize the outcome as settled under Utah-specific AI citation precedent without checking current appellate and disciplinary decisions.

The decisive question is not "Who used ChatGPT?" It is "What did the attorney do to verify the authorities before putting their signature on the filing?"

ChatGPT can make mistakes. Check important info.
Latest response




Thinking effort
Medium