Courts are increasingly taking action against attorneys who submit poorly vetted AI-generated legal briefs riddled with fabricated case citations and misquoted rulings. A Louisiana district court judge sanctioned a lawyer last month after discovering seven invented quotes falsely attributed to previous decisions in a personal injury case. The attorney received a US$1,000 (RM4,047) fine and was ordered to complete a three-hour course on responsible AI use in legal practice.
The problem extends far beyond isolated incidents. French lawyer Damien Charlotin has documented 1,600 instances of AI hallucinations in court filings from 35 countries since April 2023. The United States accounts for the vast majority at 1,116 cases, followed by Canada with 173, Australia with 74, and Britain with 59. These errors range from fabricated legal precedents and plagiarised arguments to outdated legal guidance, with most instances involving mainstream chatbots like OpenAI's ChatGPT rather than specialised legal software.
Judges have been explicit about the consequences of such negligence. New York's Supreme Court Justice Linda Kevins warned that AI-generated errors waste resources for opposing parties and the court itself, while damaging the reputation of judges whose names appear in false citations. Federal judge Sharion Aycock recently sanctioned lawyers on both sides of a Mississippi civil case for submitting briefs citing non-existent cases. Legal experts contend that ignorance of AI limitations is no longer acceptable, with American Bar Association task force member Scott Schlegel bluntly stating that attorneys must simply perform their fundamental duty to verify sources.
The underlying cause lies in how artificial intelligence operates. These systems excel at pattern recognition and can easily generate fabricated legal citations that follow standard formatting conventions, making them appear authentic. Because no lawyer can memorise every past ruling, the only reliable method involves manually verifying each citation. Yet many practitioners continue failing to do so, compounded by time pressures that force rushed submissions.
Hallucinations in court documents have surged eightfold over the past year, signalling an urgent need for better verification practices. While Charlotin is developing an AI tool to help lawyers detect errors before filing, he acknowledges it cannot guarantee accuracy. Ultimately, he emphasises that lawyers must adapt their review processes accordingly. Some courts are maintaining growing registries of AI-related calamities, and judges show no sign of softening their stance against careless filings.
