On Tuesday, May 26, 2026, CalMatters reported that two of California’s largest courts are piloting an AI tool designed to assist judges by drafting orders and generating research memos.

The Los Angeles County Superior Court and Riverside County Superior Court are testing this technology, developed by the company Learned Hand, which utilizes language models from Anthropic, OpenAI, and Google. While currently being used primarily for civil cases, there’s a possibility of expanded applications into criminal and family law, where the stakes are significantly higher.

The Los Angeles County Superior Court has a contract worth approximately $314,000 for a pilot program that began in February. This contract includes a roadmap to assess the tool’s utility in criminal, family, and probate divisions. Officials have not detailed the specific criteria for evaluating whether the AI can be safely deployed in these more sensitive areas.

Riverside County has a $10,000 agreement to test the program, with civil and probate attorneys using the tool to draft research memos, a task traditionally performed by research attorneys.

Despite the potential for increased efficiency and reduced court backlogs, concerns have been raised by legal professionals. One judge, speaking anonymously, expressed alarm at the prospect of the AI being used to evaluate appeals related to racial bias claims, particularly in light of the Racial Justice Act passed in 2020. This judge stated, “AI cannot and never will be able to replace human judgment in evaluating complex social dynamics. Ultimately, that will erode the public’s confidence in the competence and fairness of the judiciary.”

The use of AI in the judicial system is not without precedent for errors. Researcher Damien Charlotin has documented hundreds of instances where litigants, lawyers, and judges have made mistakes using AI, including nearly 90 cases in California state or federal courts since August 2024. A notable case involved a Los Angeles-based lawyer receiving a $10,000 fine for citing non-existent cases. Furthermore, a 2025 MIT study suggests that large language models can reduce critical thinking and brain activity.

Learned Hand’s CEO, Shlomo Klapper, acknowledges that AI machines can be biased but emphasizes that their tool can be tested and that this is a benefit over human fallibility. He believes AI is essential for judges and staff attorneys to keep pace with the increasing caseload. However, public defenders have echoed concerns about AI’s ability to understand the nuances of cases involving racial undertones, arguing that such language is beyond the current capabilities of artificial intelligence.

While both courts have stated that testing is being conducted on decided motions and not on live cases, the contracts do allow for testing on ongoing cases. The Los Angeles Superior Court spokesperson confirmed that while the contract permits the tool’s use for Racial Justice Act petitions, this possibility has not yet commenced. The court leadership is proceeding cautiously, with evaluations on a quarterly basis to determine the tool’s effectiveness before extending its use beyond civil cases. The Judicial Technology Advisory Committee chair, Judge Samantha Jessner, believes there is a duty to explore AI’s potential in the judicial branch.

Despite the ongoing testing and potential benefits, some, like Los Angeles County District Attorney Nathan Hochman, remain hesitant to trust AI with critical decisions, especially those involving an individual’s liberty. Hochman stated, “It’s the analysis of the case itself, coupled with the conclusions that will be reached, that I’m very hesitant to trust AI at this point in large part, because I don’t know all of the inputs that AI is using to make its decision.”

 

 

Source: CalMatters