On July 23, 2025, U.S. District Judge Julien Xavier Neals of the District of New Jersey withdrew a previously issued opinion in In re CorMedix Inc. Securities Litigation, No. 2:21-cv-14020, after defense counsel identified significant factual and citation errors. The mistakes—misstated case outcomes, non-existent quotations, and inaccurate attributions—closely resembled the kinds of issues commonly associated with so-called “hallucinations” produced by generative artificial intelligence.
In a letter submitted the day before the withdrawal, Andrew Lichtman, a partner at Willkie Farr & Gallagher and counsel for the defendants, alerted the court that the opinion had relied on three decisions to support denying a motion to dismiss, though each of those cases had actually resulted in dismissal. The letter also noted that multiple quotations cited in the opinion did not appear in the referenced decisions and that certain statements were erroneously attributed to the defendants despite not appearing in the complaint.
Judge Neals responded by entering a docket order withdrawing the opinion and accompanying order, stating that they had been entered in error. No revised opinion has yet been filed.
Although there is no confirmation that artificial intelligence played a role in drafting the withdrawn opinion, the nature of the errors has prompted discussion within the legal community, particularly given the judiciary’s recent scrutiny of similar lapses in lawyer filings. The episode serves as a reminder that the obligation to ensure accuracy in legal writing applies across the bench and bar—regardless of whether the source is human or machine.
