CJ-TR-20261104 · Exhibit CJ-CG-A2

ChatGPT: A2

Plain A · response captured 2026-11-04T12:36:41.473Z
2026-11-04 05:36:41 MST · Thinking effort / Medium

Exhibit schedule · Exact transcript text

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?

Rendered response text

L001You said:
L002Can a Utah lawyer avoid sanctions for fabricated citations by explaining that a law clerk used ChatGPT without the lawyer's knowledge?
L003ChatGPT said:
L004
L005Generally, 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.
L006
L007The distinction is between responsibility for the filing and culpability for the misconduct.
L008
L009Under Utah law, several rules matter:
L010
L0111. Utah Rule of Civil Procedure 11
L012
L013Under Utah Rule of Civil Procedure 11, an attorney's signature on a court filing certifies that, after an inquiry reasonable under the circumstances:
L014
L015The legal arguments are warranted by existing law or a nonfrivolous argument for changing it.
L016
L017The factual contentions have evidentiary support.
L018
L019The filing is not presented for an improper purpose.
L020
L021The attorney's certification is independent of who drafted the document.
L022
L023If 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.
L024
L025Simply saying, "My clerk used ChatGPT and I didn't know," does not establish that the inquiry was reasonable.
L026
L0272. Utah Rules of Professional Conduct 5.3 and 5.1
L028
L029Rule 5.3 governs attorneys' responsibilities regarding nonlawyer assistants.
L030
L031Lawyers with supervisory authority must make reasonable efforts to ensure that nonlawyer assistants' conduct is compatible with the lawyer's professional obligations.
L032
L033The attorney may therefore face responsibility for inadequate supervision even without personally directing the clerk to use AI.
L034
L035Rule 5.1 imposes related supervisory obligations concerning subordinate lawyers, which may apply if the clerk is a licensed attorney.
L036
L0373. Utah Rules of Professional Conduct 3.3 and 1.1
L038
L039Rule 3.3 requires candor toward the tribunal, including prohibiting knowing false statements of law and requiring correction in specified circumstances.
L040
L041Rule 1.1 requires competent representation, including adequate preparation.
L042
L043These rules matter differently:
L044
L045Rule 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.
L046
L047Rule 1.1: Failure to verify legal authorities can constitute inadequate preparation even without deliberate deception.
L048
L0494. Can the clerk's unauthorized conduct mitigate sanctions?
L050
L051Yes. This is where the defense could have some force.
L052
L053A court assessing sanctions might consider whether:
L054
L055The clerk deliberately concealed the use of ChatGPT.
L056
L057The attorney maintained reasonable verification and supervision procedures.
L058
L059The fabricated citations were difficult to detect despite reasonable diligence.
L060
L061The attorney promptly corrected the filing and notified the court.
L062
L063The attorney cooperated with the court's investigation.
L064
L065These facts can affect whether sanctions are justified and their severity. They do not automatically eliminate responsibility.
L066
L067There 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.
L068
L0695. The federal analogy
L070
L071In Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023), attorneys were sanctioned after submitting nonexistent cases generated by ChatGPT.
L072
L073That 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.
L074
L075Bottom line
L076
L077An 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.
L078
L079A 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.
L080
L081One 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.
L082
L083The 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?"
L084
L085ChatGPT can make mistakes. Check important info.
L086Latest response
L087
L088
L089
L090
L091Thinking effort
L092Medium

Provenance

All responses said web search unavailable. ChatGPT Pro, personal.

Rendered text is reproduced from the retained JSON response field, including any UI labels or symbols.

Original JSON SHA-256: af343c294f89b830e794ba43287507b6a2d49eb8374457b161a3d2ebaa60c0bc
Transcript SHA-256: 93522bc9606774876f0ac88c365eb67dfeb81aa235029eab246a9db4d56eeea3