On July 16, Chief U.S. District Judge Hala Y. Jarbou from the Western District of Michigan expressed concerns over a legal filing by the Department of Justice (DOJ), which potentially used a nonexistent court decision generated by artificial intelligence. This arose during a habeas corpus case involving an Immigration and Customs Enforcement (ICE) detainee, Izzeddin Ahmed Abdulgahffar Daghra. The case, which challenged an automatic government stay of a bond order, was eventually dismissed as the stay expired and Daghra was set to be released after posting a $35,000 bond.

Judge Jarbou, appointed by President Trump, highlighted the issue in a 3-page order, specifically addressing the problematic citation. The DOJ claimed the Sixth Circuit had ruled against reevaluating evidence underlying bond decisions or questioning Immigration Judges’ discretion, referencing Taylor v. Hott, a case that the judge confirmed does not exist.

This incident reflects ongoing tensions between the judiciary and the Trump administration, amidst increased scrutiny from federal judges towards government actions. The judge emphasized that attorneys utilizing AI must rigorously verify their work before submitting legal documents. She indicated that AI-generated legal research has led to multiple cases with fabricated citations, urging vigilance against briefs that include nonexistent cases.

Though she chose not to impose sanctions, Judge Jarbou insisted that future government filings must be free of false legal references. A similar issue occurred in North Carolina in April when U.S. Magistrate Judge Robert T. Numbers II reprimanded former federal prosecutor Rudy Renfer. He had used generative AI in a court brief, leading to false quotations and inaccurate citations.