In an order that noted an attorney's remorse, a Connecticut federal judge sanctioned a solo practitioner $500 this week for submitting a brief packed with false, AI-generated case citations, finding the fake authorities wasted court resources, risked misleading a pro se litigant and undermined trust in the judicial system.

In her Monday order, U.S. District Judge Janet C. Hall found that attorney David P. Stich had relied on a program called Descrybe.AI in drafting an opposition brief on behalf of his client, Milton Robinson Teletor Cojom, who is pursuing wage claims against a New Canaan, Connecticut, pizza parlor.

The order notes that AI tools have entered the mainstream of the legal profession, pointing out that both Westlaw and LexisNexis offer artificial intelligence-fueled tools to subscribers. Therefore, Judge Hall said, courts should work to direct lawyers' responsible use of the technology, rather than banning it altogether.

"This order should not be construed as a Luddite attack on technology and the efficiency it brings to the legal profession," Judge Hall said. "Rather, this order is an acknowledgement that AI remains a nascent technology with questionable reliability at this juncture.

"Given the ethical obligations lawyers must honor, it is imperative that lawyers use AI with diligence and care," Judge Hall went on. "This technology is too unsophisticated and must necessarily yield to a lawyer's obligation of candor to the court."

Upon realizing he had included numerous citation errors in a brief in July, Stich filed a declaration in which he expressed "sincere and unreserved apology" for allowing the errors, the judge noted.

In September, Judge Hall warned Stich that she might be compelled to issue an "eye-catching sanction" in an effort to dissuade other attorneys from making similar errors.

Stich "committed 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," Judge Hall said in her order.

By way of explanation, Stich said he learned of the tool from his attorney daughter, who he said touted its effectiveness, Judge Hall noted. As soon as he learned of the erroneous citations, Stich said, he canceled his paid subscription and committed to no longer using artificial intelligence. Stich said that, though he is a solo practitioner, no one who works for his practice in the future will be permitted to use AI.

The judge noted he also took it upon himself to take continuing legal education courses on the use of and risks associated with AI.

In the suit, Stich represents an employee pursuing claims against Roblen LLC and owner Viktor Berisha, alleging violations of the Fair Labor Standards Act and Connecticut's Minimum Wage Act.

Berisha failed to appear, and Judge Hall issued a default judgment in Cojom's favor, awarding more than $214,000 in damages, $12,600 in attorney fees and $758 in costs. Berisha then made an appearance and moved to set aside the judgment.

It was in Cojom's opposition to Berisha's motion to set aside judgment that Stich inserted the hallucinated citations.

"The danger of attorney Stich's AI use is especially felt here because his opponent's pro se status meant that there was not an adversary capable of calling the attention of the court to the phony citations," the judge stated in her Monday order.

Judge Hall denied Berisha's motion in July before ordering Stich to show cause why he should not be sanctioned.

"Judge Hall's decision was well reasoned and thoughtful," Stich said in an email to Law360 on Tuesday. "I am happy to have this matter behind me."

Berisha did not immediately respond to a request for comment Tuesday.

Cojom is represented by Richard Rapice of the Law Office of Richard J. Rapice LLC, and David P. Stich.

Berisha is representing himself.

The case is Cojom v. Roblen LLC et al., case number 3:23-cv-01669, in the U.S. District Court for the District of Connecticut.