Paul Felix, a judge in New York, found a court reporter had inserted AI-generated mistakes into an official transcript and issued a warning to all stenographers about their use of the technology.
The issue arose in a memorandum decision filed on July 23 regarding a drug sale case where the buyer overdosed and died. Judge Felix noted in a footnote that the transcript contained errors typical of generative AI. Attorney Rob Freund spotted the footnote on X.
“At one point in the transcript, a motion, presumably made by the State, is attributed to the trial court. At another point, an objection, presumably made by Williams, is attributed to the Bailiff. At yet another point, the State’s closing argument is attributed to the trial court,” Judge Felix wrote. “These errors, among others not described herein, complicated but did not substantially impede our review of Williams’s appeal.”
The judge continued: “Regardless, we remind the Court Reporter that this court relies on transcripts being true and accurate representations of the transcribed proceedings. Based upon the types of errors reviewed, it appears that generative artificial intelligence may have assisted with the preparation of this transcript. While AI can improve efficiency and be a productive tool for many professionals, it is incumbent upon those using such systems to proofread and ensure the accuracy of the generated product.”
For years, judges have caught lawyers using AI in court filings. These situations are messy and embarrassing for attorneys whose job is to represent clients by citing existing case law and legal precedent, which AI routinely gets wrong. Judges are becoming more outspoken about their frustrations. In May, judges in the Supreme Court of the State of New York Appellate Division spent more than 20 minutes reprimanding several lawyers after accusing one of using AI and the others of being too sloppy to catch it. The judges called the entire situation “striking, concerning, disappointing, and saddening.” Lawyers, meanwhile, blame paralegals, head colds, and rushing.
This is the first time a court reporter has been publicly put on notice for not catching AI-generated errors in transcripts. This raises the specter of errors not just in court filings from attorneys but in the records of official proceedings of a trial. There are many apps and companies that offer AI-generated transcripts for court reporting, but stenographers say their expertise as human listeners and skilled transcribers is still extremely valuable in the courtroom.
Stenographers argue that AI tends to guess rather than pause to ask for a re-statement or resolve ambiguity before putting words into the court record. A human operator can verify what was actually said, ensuring the official record matches the proceedings.
![arXiv implements 1-year ban for papers containing incontrovertible evidence of unchecked LLM-generated errors, such as hallucinated references or results. [N]](https://ai-maestro.online/wp-content/uploads/2026/05/arxiv-implements-1-year-ban-for-papers-containing-incontrove-ai-768x768.jpg)
