Federal judges are drawing hard lines for lawyers who bring sloppy artificial intelligence work into the courtroom, handing out sanctions and blocking trials when attorneys cite cases that never existed. At the same time, many judges are quietly experimenting with the same technology in their own chambers, trying to figure out what it can and cannot do.
The tension reflects a deeper struggle playing out across the judiciary: how to harness AI's potential without letting it corrode public trust in the courts at a moment when that trust is already fragile.
Last month, a federal judge in Mississippi punished four lawyers and canceled a civil trial after both sides relied on AI-generated case citations that were fabricated. A few weeks later, a Michigan federal judge found the government had likely cited a fake case in an immigration filing. Damien Charlotin, a senior research fellow at HEC Paris, has tracked more than 1,700 cases where generative AI produced false information.
"AI has the possibility to be an incredible value adder and force maximizer in the federal judiciary," says Amy Cyphert, a West Virginia University College of Law professor. "But it also has the potential to undermine public confidence in the judiciary at a time where it's already at a historic low."
A Northwestern survey of 112 judges found 60 percent use at least one AI tool, though just over 22 percent employ them weekly or daily. Judges worry about hallucinations, but also about privacy and confidentiality risks that come with uploading sensitive court materials to AI systems.
The education gap is immediate and urgent. Magistrate Judge Maritza Braswell of Colorado, alongside Judge Scott Schlegel of Louisiana and U.S. District Judge Xavier Rodriguez, founded the Judicial AI Consortium to fill the void. "Because the technology is moving very fast, I think the tendency tends to be to start from a place of fear," Braswell says. "We've really created the space where we start from a place of knowledge."
Judge Liam O'Grady, who retired in 2023 from the Eastern District of Virginia, watched the dysfunction firsthand. Judges were intrigued by AI, he recalls, but skeptical of its reliability. Lawyers began using the tools without proper vetting, and "mistakes were being made right away."
The coalition emphasizes that the fix is not prohibition but training. Schlegel points out that courts did not ban Zoom when a lawyer showed up to a hearing with a cat filter. The real difference, he notes, lies in the stakes: when judges sign an order with an error, that becomes law.
Braswell also cautions against fixating only on hallucinated cases. Other risks loom larger in her view: algorithmic bias and whether AI tools will widen the gap between judges and litigants who can afford the technology and those who cannot.
O'Grady, who started practicing in the 1970s, frames the learning curve as daunting but necessary. "There's been such a dramatic jump in technology," he says. "It's like learning a new language when you're 60 or 70 years old. We've got a whole generation of judges who this is just completely new to, and they're cautious and skeptical."
Author James Rodriguez: "The judiciary is right to throw elbows at careless AI use, but judges need to stop acting like the technology will disappear if they ignore it long enough."
Comments