Federal Judges Admit AI Use Led to Error-Ridden Court Orders

U.S. District Judges Julien Xavier Neals and Henry T. Wingate admitted that staff members used generative artificial intelligence to prepare court orders that contained significant factual inaccuracies and serious errors. The admissions followed an inquiry by Senate Judiciary Committee Chairman Chuck Grassley after lawyers raised concerns about the error-ridden rulings.

Judge Neals, based in New Jersey, reported that a law school intern used OpenAI's ChatGPT for legal research without authorization in a securities lawsuit, violating both chamber and law school policies. Judge Wingate, based in Mississippi, stated a law clerk used Perplexity as a drafting assistant to synthesize docket information in a civil rights lawsuit. Both judges described the issuance of the documents as a lapse in human oversight, noting that the faulty papers were early drafts that bypassed standard review processes before being erroneously docketed.

In response, both judges implemented enhanced oversight measures, including mandatory independent reviews of draft opinions and written AI policies. Judge Neals' new policy explicitly prohibits the use of generative AI in legal research or the drafting of orders.

Senator Grassley commended the judges for their honesty but urged the judicial branch to establish more decisive and permanent AI guidelines. He stated that the judiciary must ensure that overreliance on artificial assistance does not compromise integrity or violate the rights of litigants.