The blame game over AI hallucinations in court filings has started

The legal sector is grappling with the repercussions of artificial intelligence errors, as attorneys begin to openly identify the software that contributes to their mistakes. A notable incident involved a Louisiana personal injury lawyer, Ross LeBlanc, who publicly acknowledged errors in court briefs that cited nonexistent quotes after using an AI tool named Eve for drafting. In a letter to Judge William Jorden, LeBlanc expressed his confusion over how he could have made such a blunder. Initially, he had trusted the AI's accuracy, often finding the citations to be correct, which led him to stop double-checking the information. This misjudgment resulted in significant embarrassment when opposing counsel highlighted the inaccuracies in two filings in the 19th Judicial District Court in Baton Rouge. Eve's CEO, Jay Madheswaranm, defended the software, stating that a thorough review confirmed no inaccuracies in the case citations provided by Eve. This incident is part of a broader pattern in which law firms face sanctions for utilizing AI-generated content that includes fabricated information—often referred to as 'hallucinations' in the legal community. The spotlight is now on legal technology companies, as more attorneys name the tools they use in court documents, leading to potential reputational damage for these firms. Companies like Eve, Harvey, and Legora have heavily invested in technology that promises to enhance legal efficiency, but any slip-ups could undermine the trust that lawyers place in their capabilities. Researcher Damien Charlotin noted that less than 10% of cases disclose the AI tools involved, as many legal professionals may fear repercussions from using unverified free chatbots or other unauthorized applications. Previous high-profile cases have also highlighted the risks, including a Latham & Watkins attorney who mistakenly cited a non-existent article after relying on Anthropic's AI, Claude. Eve, valued at $1 billion after a significant funding round, is designed to assist lawyers in various tasks, from drafting documents to managing discovery requests. However, the responsibility for the final output still lies with the human attorney, which Madheswaranm emphasizes in the company’s contracts. As the legal industry continues to embrace AI technology, the likelihood of errors may rise. Courts are increasingly aware of these tools, prompting lawyers to scrutinize filings for inaccuracies that could damage their opponents' credibility. Chad Dudley, a partner at Dudley DeBosier, insists on thorough training for attorneys to ensure responsible use of AI, while LeBlanc has expressed a desire for colleagues to learn from his experience. He acknowledges the benefits of AI but feels a need for a pause to reassess its use in his practice. The ongoing struggle for accountability in the intersection of law and AI highlights the pressing need for vigilance and responsibility as technology becomes an integral part of legal processes.