A Los Angeles law firm defending State Farm in a contentious home-repair lawsuit has admitted that artificial intelligence tools were used to draft court filings containing fabricated case citations, invented quotes and legal holdings that simply don’t exist.
The admission came after attorneys representing former Carson homeowner Fa’alagilagi Meni-Siliga flagged the errors in motions submitted by State Farm’s legal team. In a court filing, attorney Eric Khodadian, who represents Meni-Siliga, said he uncovered numerous instances of what he called AI “hallucinations” — false legal references that appeared authoritative but had no basis in actual case law.
Meni-Siliga, a U.S. Postal Service letter carrier, sued State Farm along with a contractor, the contractor’s insurer and a public adjuster in July 2024 over the botched rebuilding of her home following a 2020 fire. According to the lawsuit, the blaze left her Carson house uninhabitable, and while repairs were underway, a subsequent storm caused additional water damage. She says delays in getting State Farm’s sign-off on repair work meant the job was never finished.
The fallout, according to court documents filed in January, was devastating. Meni-Siliga and her husband exhausted their savings and retirement funds, filed for bankruptcy, and ultimately lost the home to foreclosure.
“My family and I have lost forever our family home,” she wrote in the filing. “We do not come from money … It took many years and sacrifice for our family to purchase our home.”
State Farm had initially responded to the lawsuit by arguing that negligence on the part of Meni-Siliga, her contractor and the public adjuster was to blame for her losses, and separately contended she missed the legal deadline to file her claim. The case is scheduled for trial in October.
The dispute over AI-generated errors surfaced publicly during an Aug. 7 case management conference in Los Angeles Superior Court, where Khodadian presented his findings. In a subsequent filing, he described a tense scene afterward, saying a State Farm attorney appeared “visibly enraged” and followed him and his co-counsel down a courthouse hallway, repeatedly demanding to see the list of errors they had compiled.
That attorney, Kenneth Katel of the firm Musick, Peeler & Garrett, later filed a formal apology to the court and to Meni-Siliga. Katel said a colleague on the case had used AI tools to help prepare some of the filings without his knowledge. “I was not aware that AI had been used, but as lead trial counsel I accept full responsibility for our filings,” he wrote, adding that the firm has since revised its internal policy on AI use, though he did not detail the changes. Katel pushed back on Khodadian’s account of the hallway confrontation but acknowledged, “it is accurate to say I was upset given the seriousness of the allegations.”
Both Khodadian and Katel declined to comment further when contacted.
The attorney who relied on AI, Jacquelene Robinson, explained in her own filing that she discovered seven case citations across eight separate filings that were entirely fictitious, along with several mistaken case titles and quotations that didn’t actually appear in the cases she referenced. Robinson said she had used an AI research tool called Irys, mistakenly believing it was integrated with her firm’s subscription to Westlaw, a widely used legal research platform, and that it automatically verified citations. That assumption, she acknowledged, was wrong.
A State Farm spokesperson, Tom Hartmann, said the company is reviewing the matter. “State Farm expects its external counsel to conform to the highest level of ethical standards and professionalism, including confirming the accuracy of all legal filings,” he said. Hartmann did not respond to questions about how many outside law firms represent State Farm or whether the insurer vets firms’ AI policies before retaining them.
The episode reflects a growing problem across the legal profession as AI tools become more common in casework. Roughly 41% of law firms and 47% of corporate legal departments now report using generative AI in some capacity, according to a recent industry survey. Stanford law professor Daniel Ho, who studies the issue, said fabricated legal citations — often called hallucinations — have become increasingly common as AI adoption spreads. In research he co-authored last year, Ho found that some legal AI companies had overstated their tools’ ability to produce hallucination-free citations.
A global database tracking AI-related legal errors, maintained by a Paris-based researcher, has logged nearly 1,922 documented cases so far, the majority of them in the United States. Last year, a California attorney was fined a record $10,000 after submitting a legal brief in which 21 of 23 quotations turned out to be fabricated by ChatGPT.
The Meni-Siliga case now adds to a growing list of cautionary tales illustrating the risks of unchecked AI use in the courtroom — and the potential consequences for both attorneys and the clients they represent.
Original source: CalMatters
