Lawyer Cites ChatGPT-Invented Fake Witnesses in Murder Appeal

AI-generated legal briefs can now produce fabricated evidence, risking disciplinary action and invalidating case filings.

New Mexico attorney Stephen D. Aarons has been sanctioned $5,000 and referred to the Disciplinary Board after using ChatGPT to generate a legal brief that included fabricated witnesses and testimony. Aarons was representing a man convicted of murder, and his fabricated submissions were discovered by judges during a hearing. The judges expressed dismay at Aarons' reliance on AI, noting that the issue of AI "hallucinations" in legal filings is widely publicized. Aarons admitted to his "stupidity" and stated he "assumed" ChatGPT would provide a reliable summary. He was barred from appearing before the court pending the disciplinary board's investigation and was removed from his client's case, with a public defender assigned.

This incident highlights the significant risks legal professionals face when relying on generative AI without rigorous verification. The court's strong rebuke and Aarons' personal sanction underscore that AI output is not a substitute for due diligence. It suggests that the legal profession must rapidly establish clear guidelines and best practices for AI tool usage to maintain the integrity of the judicial system.

The widespread awareness of AI "hallucinations," as noted by the judges, implies a growing disconnect between the capabilities of AI and the professional responsibility of lawyers. This case signals a potential shift in judicial scrutiny, with courts likely to become more vigilant in detecting AI-generated inaccuracies and holding practitioners accountable for submitted falsehoods, regardless of their origin.