Header Photo Example photo of aging labrador – we couldn’t find Kyra
A California custody lawsuit over an aging Labrador ended with both sides being punished for citing nonexistent precedents faked by AI
The seemingly limitless proliferation of cases in which lawyers have been caught letting fictitious AI-generated legal citations contaminate their briefs continues to amaze.
That’s not only because judges are fining more lawyers for their laziness, but because the publicity about these embarrassments has been inescapable.
Here’s one involving a dog named Kyra.
She’s a 16-year-old Labrador retriever who became the target of a nasty custody fight between a California couple after the dissolution of their domestic partnership. In the course of the lawsuit, one lawyer published two AI-fabricated citations in a filing. The opposing law firm didn’t catch the flaw and cited the same fake cases in its filings, including in a court order signed by a judge.
Most lawyers grew up in a time when you could expect the other side to spin and even to lie about the record some of the time, but just lying or making a mistake about the existence of a case was basically unheard of up until a few years ago.
— Eugene Volokh, UCLA law school
The case of Joan Pablo Torres Campos vs. Leslie Ann Munoz also points to how AI, touted worldwide as a labor-saving technology, has actually increased the workload in some trades and professions, like lawyering. For litigators, it has created a new imperative: ferreting out citations that have been fabricated by AI bots in their own court filings — and their adversaries’.
I’ve written before about the proliferation of AI-generated fabrications infiltrating legal filings and even legal rulings, despite the advice drilled into the heads of even law students about making sure that their citations to precedential cases are accurate. But the wave keeps building: A database of AI hallucinations maintained by the French researcher Damien Charlotin now numbers 1,174 cases, of which some 750 are from U.S. courts
That’s almost certainly a conservative count. Most AI fabrications may not even come to the attention of litigants or judges, especially in state courts.
“For every case that talks about this, my guess is that there are many that aren’t visible,” says Eugene Volokh of UCLA law school and the Hoover Institution, who keeps a weather eye on AI-related courthouse developments. He believes there may be thousands escaping notice.
AI has introduced mistakes that were never seen in the past. “Most lawyers grew up in a time when you could expect the other side to spin and even to lie about the record some of the time, but just lying or making a mistake about the existence of a case was basically unheard of up until a few years ago,” Volokh told me. “That’s because there would be no source of hallucinations — maybe you’d get the citations slightly wrong or you mischaracterized or misquoted them, but to talk about a case that doesn’t exist — that didn’t happen. Now it happens a lot.”
The judiciary is getting increasingly nervous about AI fabrications becoming part of the judicial record. “Reliance on fake cases…seriously undermines the integrity of the outcome and erodes public confidence in our judicial system,” an appelate judge stated.
Read full article
