Law360 (June 24, 2026, 12:55 PM EDT) -- The rules surrounding artificial intelligence experimentation in courts run the gamut from court systems offering proprietary tools and training to unwritten policies that essentially amount to don't ask, don't tell.

With no overarching guidance, each federal district and each state court system is on its own to develop internal AI policies.

"There are no standards right now. We're all operating on an ad hoc basis in the federal judiciary," said U.S. District Judge Xavier Rodriguez in the Western District of Texas. "Nothing has been developed. There's 94 judicial districts in the United States federal system … we've got basically 94 independent IT departments, and so everybody is struggling to figure this AI thing out."

Courts may be drawn to use AI to handle simple tasks such as scheduling, which consume a lot of judges' time, according to retired federal judge Katherine B. Forrest, now the co-chair of Paul Weiss Rifkind Wharton & Garrison LLP's global artificial intelligence group. Another common use is combing through mounds of documents to create summaries that help judges and clerks as they review filings, or compiling a list of questions judges may use for oral arguments.

Sometimes, however, lack of funding for vetted AI tools or for dedicated IT staff stand in the way of many courts approving judicial use of the technology, Forrest said.

"There's incredibly high variability as to what tools judges have access to and what kind of training they have across the country," she said. Though some judges are still exploring publicly available tools such as ChatGPT, Forrest said it was her impression that the number was "a minority, and it's shrinking."

Here, Law360 offers three snapshots as a window into the way various courts are handling the new technology at this time.

New Jersey at the Fore

The New Jersey Courts system has been exploring the use of generative AI since 2023, according to Judge Michael J. Blee, the system's administrative director. Early on, the court system began collaborating with the New Jersey State Bar Association to produce guidance and best practices for integrating AI into the courts. It also built in data protections.

"We wanted to build a secure internal AI tool," Judge Blee said. "So, it's built on a public LLM [large language model] engine, but it's fully protected within the judiciary infrastructure … Only authenticated judiciary users have access, and the way it was designed was done for uploading and querying internal documents."

Judge Blee said that when AI was first introduced, most judges' initial reaction was "fear," but given time, the apprehension has evolved into excitement.

Training has been offered since the very beginning, according to Jessica Lewis Kelly, special assistant to the administrative director of the New Jersey Courts. From a pilot program of just 11 participants in early 2025, the AI tool now has more than 2,000 individual users from courts across the state, Lewis Kelly said.

Every user is invited to attend a virtual training, and court administrators regularly travel throughout the state to offer in-person training.

Bar association collaboration means that judges and attorneys are on the same page with ethical AI use, and the courts offer continuing legal education credits for attending AI training, Judge Blee said.

"I do think it is unique among other state court systems, because we have engaged our judges from the very, very beginning, and it's their feedback and their input that has enabled the tool to evolve and also has enabled this ongoing close relationship with the Bar Association," Lewis Kelly said.

Innovation in San Antonio Through Coordination

Collaboration is not so easy to come by in federal district courts, which are "kind of siloed," Judge Rodriguez of the Western District of Texas said.

But just because federal courts don't formally integrate doesn't mean they are blocked from sharing resources. The judge is working toward developing an industry-specific generative AI tool for use in the Western District of Texas, which could one day be adopted by other federal courts.

Beyond presiding over his San Antonio courtroom, Judge Rodriguez is building a community of judges intent on developing and sharing knowledge of AI tools.

In January, Judge Rodriguez became a founding member of the Judicial Artificial Intelligence Consortium, which is also headed up by U.S. Magistrate Judge Maritza Dominguez Braswell of the District of Colorado and Judge Scott U. Schlegel of the Louisiana Fifth Circuit Court of Appeal.

At last count, the consortium numbers 325 members representing courts of all sizes around the country, Judge Rodriguez said.

"It's a judges-only platform, and we have, on sort of a quarterly basis, just an open forum where judges can start asking these questions, because, like I said, no one's providing us any guidance," Judge Rodriguez said. "We've kept it free from vendors. It's just judges asking other judges questions and seeking help and ideas. We're all trying to navigate this on our own."

Judge Rodriguez said he is following best practices for the use of AI in some of the same ways Forrest described, and emphasized he does not use artificial intelligence for any adjudicative tasks.

Cook County Slow to Launch

Some courts have yet to formally acknowledge judicial AI use. That includes Cook County Judicial Circuit Court in Illinois, where Judge Nichole Patton said judicial officers without AI training or resources are dealing with a rise in drafts that are believed to be AI generated.

"My colleagues are informing me that they're seeing errors or facts in the case that really don't exist, things of that nature," Judge Patton said, noting that the issues mostly arise among pro se litigants, of whom she sees few in her law division-trial section courtroom.

AI filings from self-represented litigants can be "kind of obvious," Judge Patton said, drawing concern from her judicial colleagues who lack support from law clerks to thoroughly review the documents.

"We're getting inundated with this paperwork," Judge Patton said.

Judge Patton, who served as president of the Chicago Bar Association for bar year 2025-2026, declared that AI was the theme of the year for the organization.

But even as the Chicago Bar is offering resources for members to learn about AI, judges in the local circuit have not yet been cleared to use the tools themselves.

"We haven't been presented with the opportunity to have the discussion about AI, and the effects of AI," Judge Patton said. "At this point, we don't know if any of the judges are using AI. If so, what are they using? We have no clue."

And, Judge Patton said, even if judges are given the opportunity, apprehension remains.

"I have more education on AI than most, because it's my presidential theme, and this is what I've been working on all year, but a lot of judges are still reserved," Judge Patton said. "They keep hearing about the hallucinations, and they're like, 'Oh, you can't trust AI,' and, 'That's cheating,' and, 'You know, AI can't do work for you.' So, we're so inundated with the negative."

"We pride ourselves on saying, 'I don't use AI. I read my cases' … AI is not here to replace us, it's here to assist us, just like you have a law clerk," Judge Patton said.

--Editing by Nicole Bleier.

For a reprint of this article, please contact reprints@law360.com.