By Lauren Berg

Law360 (December 12, 2023, 11:57 PM EST) -- A New York federal judge on Tuesday ordered one of Michael Cohen's attorneys to provide copies of three apparently fictional cases cited in a recent motion, in an incident appearing to echo other instances in which attorneys have been tripped up by the tendency of artificial intelligence to "hallucinate" legal citations.

U.S. District Judge Jesse M. Furman ordered Cohen's criminal defense attorney, David M. Schwartz of Gerstman Schwartz LLP, to provide copies of three decisions cited in a Nov. 29 motion to discharge Donald Trump's longtime fixer from supervised release or explain why he should not be sanctioned, as was the recent fate of two attorneys who submitted a brief written by artificial intelligence that cited nonexistent case law.

"As far as the court can tell, none of these cases exist," Judge Furman wrote, noting that one citation "refers to a page in the middle of a Fourth Circuit decision that has nothing to do with supervised release," while another "appears to correspond to nothing at all."

If he submits a show cause, the judge told Schwartz to include a "thorough explanation of how the motion came to cite cases that do not exist and what role, if any, Mr. Cohen played in drafting or reviewing the motion before it was filed," according to the order.

Judge Furman is not the only person who wasn't able to find the decisions cited in the initial motion.

E. Danya Perry of Perry Law — who represents Cohen in civil litigation, including for his testimony in the New York attorney general's civil fraud case against Donald Trump — made an appearance in the instant case just a week ago, filing her own reply in support of the motion to terminate her client's supervised release.

Cohen, who was Trump's personal lawyer and self-described "fixer" for more than a decade, was sentenced in 2018 for a slew of crimes, including tax evasion and paying off women who claimed they had affairs with Trump.

Perry argued in her letter that Cohen "has provided significant information to authorities in connection with numerous government investigations" and that other courts have granted such motions in similar situations.

The government opposes Cohen's motion, saying in its Dec. 4 brief that the disbarred attorney continues to deny responsibility for his criminal conduct and has "attempted to walk back his guilty plea to tax evasion."

But Perry, who included several of her own citations in her letter, added in a brief footnote that she was "unable to verify" several cases cited in the initial motion filed by Cohen's other counsel.

Perry told Law360 late Tuesday that she came into Cohen's criminal case after Schwartz filed the initial motion and prosecutors filed their opposition, saying she believed the government had mischaracterized Cohen's testimony in Trump's fraud trial.

"In conducting my own research in support of Mr. Cohen's motion, I was unable to verify the case law that had been submitted by previous counsel in his initial papers," Perry said. "Consistent with my ethical obligation of candor to the court, I advised Judge Furman of this issue."

"Based on the many cases I cited in my papers, I believe the motion for early termination of Mr. Cohen's supervised release to be a meritorious one," she added.

A representative for the government declined to comment.

Schwartz did not immediately respond to a request for comment.

The Courts and Tribunals Judiciary in the United Kingdom on Tuesday urged British judges to watch out for signs that artificial intelligence had been used to draft legal arguments, warning that chatbots often hallucinate case law and evidence that does not exist.

A handful of lawyers in the U.S. have already landed in hot water for their apparent use of AI in drafting court filings, including a pair of New York personal injury attorneys who were sanctioned in June for submitting a brief written by AI that cited nonexistent case law. The judge said the lawyers "abandoned their responsibilities" to check their work, and their behavior rose to "bad faith" when they then waited weeks to finally come clean about the embarrassing episode.

The attorneys apologized to seven federal and state judges and a client as part of their sanctions for submitting the brief.

The Tenth Court of Appeals in Waco, Texas, in July overruled an appellant's bid for pretrial release or a reduction to his $400,000 bail after finding that the "illogical" filing prepared by his attorney included citations to cases that don't exist.

Since ChatGPT and other AI products have exploded in popularity, some judges have created rules around the use of artificial intelligence when drafting legal filings.

U.S. District Judge Brantley Starr of the Northern District of Texas requires attorneys appearing before him to certify that they did not use generative AI to write their briefs, or that if they did, the filings were checked for accuracy "by a human being."

U.S. Court of International Trade Judge Stephen A. Vaden and U.S. District Judge Michael Baylson of the Eastern District of Pennsylvania have issued similar orders requiring attorneys and pro se litigants to disclose when they have used AI to prepare any court filings and to certify the filings' accuracy.

The government is represented by Nicolas Roos of the U.S. Attorney's Office for the Southern District of New York.

Cohen is represented by David M. Schwartz of Gerstman Schwartz LLP and E. Danya Perry of Perry Law.

The case is U.S. v. Michael Cohen, case number 1:18-cr-00602, in the U.S. District Court for the Southern District of New York.

--Additional reporting by Ryan Boysen, Henrik Nilsson, Frank G. Runyeon, Hanna Vioque, Jack Karp, Sarah Martinson and Hailey Konnath. Editing by Michael Watanabe.

For a reprint of this article, please contact reprints@law360.com.