NM Supreme Court punishes lawyer for submitting ChatGPT fabrications
Appeals lawyer fined $5,000 after admitting to using ChatGPT to defend his client, though he denies he was aware AI could fabricate witnesses or testimony.
Digital Assignment Editor
A Santa Fe-based lawyer in a high-profile appeal was found in contempt of court this week for submitting witnesses and testimony that was discovered to have been wholly fabricated by artificial intelligence.
The New Mexico Supreme Court also fined Stephen Aarons $5,000, referred him to a disciplinary board and barred Aarons from appearing before the Supreme Court pending the outcome of investigations and proceedings, court filings show.
The court said Aarons “demonstrated a lack of remorse and a lack of concern for his client,” who the court said was harmed more by Aarons’ actions than himself.
Aarons admitted to using ChatGPT in the preparation of the brief he filed with the court in his handling of the appeal of Oscar Renee Sandoval, who pleaded not guilty to the murder of his partner and the mother of his children.
In his response, the lawyer claimed he didn’t “appreciate” that ChatGPT could fabricate “witnesses, testimony, quotations, and authorities in a form that appeared coherent and plausible.”
But he also admitted to the court that he failed to verify factual claims in the AI-generated brief before signing and filing.
Such AI fabrication is widely referred to as AI “hallucination,” in which large language models like ChatGPT produce legible text that is nonetheless factually inaccurate.
In this case, ChatGPT was found to have produced “false testimony from wholly fabricated witnesses — Officer Michelle Amarillo, Officer Sanchez, Manal Al-Jibury, and Teresa Marquez,” as well as false testimony from actual witnesses in the case.
OpenAI, which develops and offers ChatGPT services in free and paid tiers online, has not responded to requests for comment.
