Lawyer Fined $5K Over AI-Hallucinated Witnesses

The New Mexico Supreme Court has fined attorney Stephen Aarons $5,000 and held him in contempt for submitting an appeal that included AI-fabricated witnesses and fake police testimony, according to Reuters. The case, which involved an appeal of his client's murder conviction, ranks among the most severe sanctions yet imposed on a lawyer for failing to verify AI-generated legal content. The fabricated material allegedly appeared in court filings as purported witness statements and police accounts that did not exist. Courts across the United States have issued repeated warnings as lawyers submit briefs containing hallucinated citations, invented case law, and fabricated facts—errors that large language models can produce with convincing fluency. What makes this ruling notable is the severity of the response. Earlier incidents often resulted in reprimands, mandatory training, or modest penalties. A $5,000 fine combined with a contempt finding signals that judges are moving from admonishment toward meaningful punishment, treating AI-assisted filings as a matter of professional responsibility rather than a technical curiosity. The underlying legal principle is not new: attorneys are responsible for the accuracy of everything they file, regardless of how it was produced. Delegating research or drafting to an AI tool does not transfer that duty. Courts have consistently held that lawyers must independently verify citations and factual assertions before submitting them. The case also raises uncomfortable questions about access to justice. AI tools can help solo practitioners and under-resourced defendants, but only if used carefully. Sanctions like this one may discourage careless use while doing little to address the deeper pressures that lead lawyers to rely on AI in the first place. For now, the message from New Mexico is unambiguous: verify everything, or face consequences.