State Farm Lawyers Admit Faking Court Citations
State Farm's outside counsel admitted to citing seven non-existent cases in court filings, triggering an internal review and apology to the judge.
The Fake Citations
State Farm lawyers admitted they cited seven court cases that do not exist. This happened in a dispute over a fire-damaged home in Carson, according to Reason Magazine. The filings were part of a motion to limit evidence in the case.
The plaintiff’s team found the errors during a hearing on Friday, August 7, 2026. They told the court the citations were “fabricated” and possibly “hallucinations.” State Farm’s lead counsel had no idea this was happening until that moment.
The lawyer who signed the filings issued a public apology to the judge and the other side. He stated he reviewed all State Farm’s motions and replies to find the truth. He confirmed the seven fake citations were in eight different filings.
The Company Response
State Farm spokesman Tom Hartmann said the company expects strict ethical behavior from its lawyers. According to reports, State Farm requires its outside lawyers to uphold the utmost ethical and professional standards, particularly regarding AI usage and verifying the precision of all filed legal documents.
The lawyer who made the mistake said he deeply regrets his lapse in judgment. He noted that the plaintiff’s attorneys were correct about the fake cases. He promised to ensure no other AI hallucinations exist in their other documents.
This incident highlights a growing problem where software invents legal facts. Reason Magazine reported on this specific case as part of a wider look at AI in law. The report shows how easy it is for bad data to enter court records.
A Broader Pattern
This is not the only time courts might be using AI to write opinions. A partner at Lehotsky Cohn LLP ran a test on 2,250 published appellate opinions. He used a tool called Pangram to check for signs of AI authorship.
The tool found that dozens of these opinions showed signs of AI writing. Some passages were so long they stretched for pages. The researcher did not name the judges to avoid targeting specific people.
Pangram claims a very low error rate for its detection. It says the false positive rate is of just 0.This figure represents 0041%, which translates to approximately one false positive for every 24,000 documents. The researcher said he has never seen the tool give a false alarm.
The article about these findings is titled Signs of AI Authorship in Federal Appellate Opinions. It suggests that courts might be using AI to help write their decisions. The author says this could be promising if done correctly.
He argues that Good writing … is clear thinking made visible. He believes AI can sharpen thinking but warns against giving up human judgment. The piece notes that the fate of legal reason and the future of human participation in the legal system is at stake.
The author concludes that Dozens of published circuit opinions show signs of AI writing. If courts are using AI, that’s promising—and worth getting right. This suggests a need for strict rules on how judges use these new tools.
The State Farm lawyer’s apology marks the first step in fixing this specific error. The court has not yet announced any sanctions or penalties for the firm. The plaintiff’s side now waits to see how the judge handles the fake citations.