Mathieu Shapiro Analyzes the Ethical Risks of Judicial AI Hallucinations
In a recent The National Law Journal article, titled “Judge’s AI Error-Laden Ruling Faces Appellate Scrutiny. Are There Lessons for Lawyers?” Obermayer managing partner and appellate attorney Mathieu Shapiro discussed evolving ethical and practical challenges of generative AI in the courtroom. The article focuses on a pending appeal before the U.S. Court of Appeals for the Fifth Circuit involving a Mississippi federal judge whose temporary restraining order contained fabricated citations, misquoted state law, and references to parties and declarations that did not exist. The judge later acknowledged that a law clerk had used the AI tool Perplexity in drafting the order.
Commenting on the matter, Shapiro emphasized that lawyers must proactively disclose known judicial errors rather than remain silent.
“[Y]ou have to assume it will surface eventually, and anything built on a flawed order is vulnerable to being unwound later,” Shapiro stated. “Raising it isn’t comfortable; you’re effectively telling the court its own opinion contains an error it didn’t catch. But the alternative—staying quiet and hoping it holds up— is far riskier for the client.”
Shapiro also addressed the impact on confidence in the judicial process, stating:
“It’s good that the Fifth Circuit is engaging with this head-on, because what’s at stake is nothing less than litigants’ confidence that they’re getting the reasoned judgment of a fully engaged judge.”
Read the full analysis here.
