California lawyer fined $1,000 by a judge for submitting 3 nonexistent cases to the court, indicating possible use of artificial intelligence (AI) tools generating “false content.” Such issues are increasingly common in courts nowadays.
Federal Security Judge Thomas S. Hixson wrote in an order on May 6 that the court does not prohibit or oppose the use of AI-generated content in legal defense, as long as lawyers can independently verify its accuracy.
Sanctions were imposed on lawyer Jessica R. Barsotti representing plaintiff Karl Gregoire in a lawsuit against the Bay Area Rapid Transit (BART). The plaintiff, a fare inspector, was fired for refusing to receive mandatory vaccines during the COVID-19 pandemic for religious reasons.
Barsotti had previously missed several court deadlines, leading the defendant to request dismissal of the case, prohibition of evidence submission by the plaintiff, and monetary fines. In response, Barsotti submitted documents to the court to address the defendant’s motion for sanctions.
Barsotti cited 4 cases in the submitted documents, arguing that the plaintiff’s interests should not be affected by attorney’s errors; however, the court later found that 3 of these cases do not exist.
These errors could jeopardize the entire litigation. The judge issued a warning of sanctions, stating the possibility of “total or partial dismissal of the lawsuit.”
Additional sanctions against Barsotti include requiring her to notify the plaintiffs of the sanctions and undergo at least 1 hour of education on “ethical use of AI in legal practice.”
The judge also ordered the court clerk to notify the State Bar of California. The organization is responsible for regulating California’s attorneys’ licenses and disciplinary matters.
Efforts to contact Barsotti for comment were unsuccessful.
Researcher Damien Charlotin pointed out that from the second quarter of 2023 to May 19 of this year, over 1,000 cases of false content generated by general AI have been discovered in the United States, accounting for nearly 70% of such cases globally. This generated content often includes fake citations, incorrect references, distortion of facts, and outdated advice.
The capabilities of generative AI in the legal field have been under scrutiny since March 2023. The performance of OpenAI’s GPT-4 model in bar exams surpassed that of over 90% of human test takers.
According to a survey conducted in 2025 by the International Legal Technology Association of over 15,200 lawyers in more than 580 law firms globally, 88% of respondents believe that AI will have a significant impact on the legal technology industry in the next 3 to 5 years.
Kevin Snider, Chief Legal Counsel of the Pacific Justice Institute, stated via email to a media outlet, “AI may provide non-existent cases or cite real cases but give incorrect legal conclusions.”
“If false evidence generated by AI is used in court proceedings, public confidence in the judicial system will collapse,” he said.
The State Bar of California has notified 5 lawyers of disciplinary charges for improper use of AI-generated false evidence, with 2 cases approved by the Bar Association’s court and awaiting final judgment by the California Supreme Court.
As of now, no lawyer in California has faced formal punishment for submitting AI-generated content.
George Cardona, Chief Adjudication Counsel of the State Bar of California, stated in an email to a media outlet that these cases highlight the consequences of not verifying submitted materials (regardless of the tool used to generate them), which can disrupt the functioning of the judiciary.