Judges are increasingly acting as both gatekeepers and users of artificial intelligence (AI) in the courtroom: they are establishing rules for lawyers and litigants who bring AI-assisted work into court, and penalizing those who fail to verify sources or who cite AI-generated, nonexistent case law.
Why this matters
As sanctions against attorneys for irresponsible AI use rise, judges’ chambers have become the focal point of tensions between innovation and caution. Some in the judiciary view AI as a tool that can enhance the administration of justice, while others remain deeply skeptical. Amy Cyphert, professor at West Virginia University College of Law, warns that AI could both add significant value to the federal judiciary and, at the same time, further erode public confidence in courts already facing historically low trust.
Adoption rates and concerns
In a Northwestern survey of 112 judges, 60% reported having used at least one AI tool; just over 22% said they used such tools weekly or daily. Beyond the problem of AI “hallucinations” — false or fabricated outputs — judges also raised concerns about privacy and confidentiality implications stemming from using these technologies.
Education initiatives: the Judicial AI Consortium
Judges told researchers that AI adoption is outpacing formal training. To address that gap, Magistrate Judge Maritza Braswell of the U.S. District Court of Colorado co-founded the Judicial AI Consortium (JAIC) with Judge Scott Schlegel of Louisiana’s circuit court and U.S. District Judge Xavier Rodriguez. Braswell said the rapid pace of technology tends to generate fear, and JAIC aims to create a space that begins from knowledge rather than fear.
Tensions and concrete incidents
U.S. District Judge Liam O'Grady of the Eastern District of Virginia said that before his 2023 retirement judges were fascinated by AI but had serious doubts about its reliability, noting that mistakes were already appearing as lawyers began using the tools. Recent cases illustrate the risk: last month a federal judge in Mississippi sanctioned four lawyers and canceled a civil trial after both sides cited AI-generated, nonexistent cases. Damien Charlotin, a senior research fellow at HEC Paris, has identified more than 1,700 cases in which generative AI produced hallucinations.
In another instance, a federal judge in Michigan accused the government of likely citing an AI-generated case in an immigration filing. Chief Judge Hala Jarbou, a Trump appointee, did not impose sanctions but wrote that “the Government must ensure its future filings with this Court do not include nonexistent case law.”
The proposed remedy: education and prudent use
JAIC’s founders emphasized to Axios that education is the key response. Scott Schlegel noted that the courts did not abandon remote hearings because one lawyer appeared as a cat on Zoom; Rodriguez described the effort to integrate AI as “a work in progress” for both lawyers and judges. Braswell also cautioned that an excessive focus on hallucinations could divert attention from other ethical issues such as bias and access to justice.
Different standards for external and internal use
Schlegel added that judges view AI use by attorneys or pro se litigants differently than their own use within chambers. “We need to be much more thoughtful in the way that we use these tools,” he said, adding that errors in an issued order carry legal weight.
Bottom line
Judges are learning to navigate a rapidly evolving technology from positions where mistakes are especially consequential. As Liam O'Grady put it, adapting to AI is like learning a new language later in life for many on the bench — a challenge that helps explain the caution and skepticism even as courts experiment with the technology.
Further reading
- AI tells lawyers how judges are likely to rule
