Labor & Employment Law Daily Wrap Up, REMEDIES, DAMAGES—D. Conn.: Attorney sanctioned $500 for use of AI tool that generated hallucinated citations, (Nov 20, 2025)

By Kathleen Kapusta, J.D.

“This Order should not be construed as a Luddite attack on technology and the efficiency it brings to the legal profession,” but rather is “an acknowledgement that AI remains a nascent technology with questionable reliability at this juncture.”

In determining that a $500 sanction was appropriate for an attorney who, as a result of his use of an artificial intelligence based legal research tool, submitted a pleading containing “hallucinated” citations, a federal court in Connecticut noted that while he abdicated his responsibilities under FRCP 11, he acknowledged his mistake, recognized the seriousness of his misconduct, and took steps to prevent future error. “Given the ethical obligations lawyers must honor, it is imperative that lawyers use AI with diligence and care,” the court stressed, noting that this “technology is too unsophisticated and must necessarily yield to a lawyer’s obligation of candor to the court” (Cojom v. Roblen, LLC, No. 3:23-CV-1669 (JCH) (D. Conn. Nov. 17, 2025)).

The plaintiff in this action sued a pizza restaurant and its owner alleging they willfully failed to pay minimum wage and overtime in violation of the FLSA. After the defendant failed to appear, answer, or otherwise defend against the lawsuit, a default judgment was entered against it. It subsequently moved for vacatur and the plaintiff submitted a memorandum opposing the motion to set aside the default judgment.

Fabricated citations. In reviewing the memorandum, the court found several citations to legal authorities that did not exist. Believing these resulted from the use of AI, the court issued an order to show cause to the plaintiff’s attorney who, in his declaration in response, admitted to using an AI tool. He also expressed “sincere and unreserved apology” for the nonexistent citations, explained that it was his first use of an AI tool, and stated that in using the tool, he recognized certain cases it referenced from his prior exposure to vacatur motions and that this familiarity provided “a false sense of confidence in the AI tool’s accuracy.” In addition, he stressed his commitment to “ensuring that this type of error does not recur and to maintaining the highest standards of accuracy, candor, and professionalism in [his] future conduct before this Court and all others.”

At the hearing, he explained that he learned about the tool from another attorney who had been practicing for 10 years. He told the court he canceled his license for the tool, he would no longer use AI in his legal practice, and, although a sole practitioner, he would ensure that no one in his firm would ever use AI in the future. He also took CLE coursework on the subject of AI technology and its risks.

New frontier. Observing that it was issuing its order “in a proverbial new frontier as society begins to grapple with both the power and potential danger of this technology,” the judge noted that courts “around the country have begun confronting the phenomenon of so-called ‘hallucinated’ citations in court papers—that is, citations to legal authority that are either partly or wholly nonexistent.”

Mata. Quoting from the Southern District of New York’s Mata v. Avianca, Inc., decision, the court here noted that many “harms flow from the submission of fake opinions” including that the “client may be deprived of arguments based on authentic judicial precedents.” Noting that the attorney here admitted he was representing “a young, scared, and vulnerable immigrant worker” against a defaulted, pro se defendant, the court explained that the plaintiff “was deprived of arguments based on legitimate judicial precedent and his pro se opponent was forced to contend with arguments based on phony law without the benefit of legal representation.”

Further, not only did the defendant’s pro se status mean it was not capable of calling attention to the phony citations, but also the court had to expend time and resources investigating the citations. “The oversight in submitting fake citations,” it observed, “is more than just sloppy lawyering: it imperils the integrity of our judicial process.”

Fundamental alteration. “Our society sits on the precipice of rapid technological development and that the continued development of AI will fundamentally alter life as we know it,” said the court, stressing the “ethical obligations lawyers must honor.” Further, it explained, “it is imperative that lawyers use AI with diligence and care. This technology is too unsophisticated and must necessarily yield to a lawyer’s obligation of candor to the court.”

Although the court found the attorney did not conduct a reasonable inquiry into his pleading as required under Rule 11, it noted that he canceled his license to the AI research tool, resolved never to use AI again in his legal practice, promised that no lawyer in his employ would use AI in the future, and completed CLE programming related to AI in the legal profession. Balancing the seriousness of his conduct with his acknowledgment of his error and the steps he took to prevent future error, the court found $500 to be an appropriate sanction.

The case is No. 3:23-CV-1669 (JCH).

Judge: Hall, J.

Attorneys: Richard Rapice (Law Office of Richard J. Rapice) for Milton Robinson Teletor Cojom. Viktor Berisha (Law Offices of Viktor Berisha) for Roblen, LLC fdba Vicolo Pizza Restaurant and Viktor Berisha.

Companies: Roblen, LLC; Vicolo Pizza Restaurant

Cases: RemediesDamages Procedure ConnecticutNews