Ethics in Brief: Verbum Sapienti Sat
By Edward J. McIntyre
“A word to the wise is sufficient.” Perhaps not.
Again and again, the same horror story. Court filings and appellate briefs, with non-existent citations, references, quotes. The hallucinogenic product of some generative AI product or other. Now it’s even infecting the federal judiciary.
How we smirked at the misfortune of the plaintiff’s lawyer in the Avianca case in 2023. His New York state case removed to Southern District of New York. A firm with no federal case research tools. Lawyer relied on teen-age son’s advice: “ChatGPT is a new, cool research app.” Hallucinations in the lawyer’s filings—more than 6 fake cases; quotes; citations; non-existent opinions. The court stated it couldn’t find cases he cited. Lawyer went back to ChatGPT. ChatGPT doubled down. More fake citations. OSC. Sanctions. National publicity. His name prominent. Humiliation.
With the resultant publicity, you’d think: verbum sapienti sat! No way. Wash, rinse, repeat!
Since the Avianca case in June 2023, as of July 2025 there have been at least 162 cases in the United States where courts, state and federal, trial and appellate, have issued decisions, frequently with sanctions, including referrals to disciplinary authorities, because of fabricated cases; false quotes; phony citations; fabricated exhibits; misrepresented precedents—the list goes on.[1]
Nor are hallucinogenic citations limited to lawyers’ submissions to courts. False citations, it appears, work both ways.
In July 2025, a district judge in New Jersey withdrew an opinion after counsel pointed out that the court misstated case outcomes and had 5 fake quotes attributed to opinions that did not contain the quotes. That same day, a federal judge in the Southern District of Mississippi withdrew an opinion after a lawyer pointed out errors in the opinion: fabricated citations; non-existent parties to the case; false quotes. Were the trusted drafters of those judicial opinions law clerks—aspiring lawyers with cherished federal clerkships?
Rule 1.1, Comment [1] makes clear that the duties stated in our competence obligation includes “the duty to keep abreast of the changes in law and its practice, including the benefits and risks associated with relevant technology.” Rule 3.3(a)(1) prohibits knowingly making false statements of fact or law to a tribunal and failure to correct a previous false statement. Rule 3.3(a)(2) prohibits misquoting to a tribunal the language of a book, statute, decision or other authority. Rule 1.0.1(f) makes clear that a lawyer’s knowledge—for “knowingly”—may be inferred from the circumstances.
Knowing 162 instances of judicial decisions across the country—with more than a dozen in California—calling out instances where generative AI produced hallucinated content (typically false case citations, including non-existent cases) could a lawyer argue today that the lawyer did not “know” the lawyer’s court submission contained phony cases or other false citations? Or that the lawyer had stayed current about the risks of using generative AI? Could a lawyer in that situation possibly explain away a failure even to read cases cited to a court—to see if the case existed; to see if it really had the quotes the lawyer included? Would “invincible ignorance” be a cogent response to an OSC or a defense to a discipline complaint?
The many benefits of generative AI are becoming manifest: it increases efficiency; reduces cost for clients; it may “level the playing field” for smaller firms coping with overwhelming discovery; it can force lawyers to think creatively in ways they would not otherwise have done.
Generative AI’s attractiveness and benefits, however, are no substitute for our ethical obligations to courts and clients to ensure that we scrupulously review court submissions; that we know every case cited exists; every quote, accurate; every representation, an accurate statement of cited precedent. Not only do our professional responsibility obligations require this, but avoiding potential reputational damage may be a strong motivator. None of us wants to be case number 163 on the list.
[1] Damien Charlotin has gathered this data for the United States and other jurisdictions, including UK, Australia, Ireland, Canada, Germany, and others. The United States leads the pack: 162 to 92, worldwide—and counting. He lists, among other data, each case citation, so determining the lawyer involved is only a couple of clicks away. See, damiencharlotin.com.
