New Mexico Court Fines Lawyer $5,000 Over ChatGPT-Fabricated Brief

The New Mexico Supreme Court has fined a defence lawyer $5,000 for submitting an appeal brief that contained fabricated witness testimony and false police statements generated by the AI chatbot ChatGPT.

Stephen Aarons, a private attorney based in Santa Fe, was held in contempt for failing to verify the accuracy of the filing. He was representing Oscar Renee Sandoval, who was convicted of killing the mother of his children and sentenced to life in prison last year. Sandoval had pleaded not guilty, and his appeal remains pending.

Fabricated material in the filing

The court found that the brief included "false testimony from wholly fabricated witnesses," along with fictional details such as statements that the shooter was wearing dark pants and a white shirt. Last month, the court asked Aarons to explain how the material ended up in his main appeal brief.

At a hearing on 21 August, Aarons told the panel that he had uploaded a computer-generated transcript and other case materials to ChatGPT, expecting the tool to produce what he called a "bulletproof summary." The judges responded with disbelief that he was unaware of the technology's capacity for error.

Justice C Shannon Bacon questioned whether he followed current events at all, noting that problems arising from lawyers relying on AI hallucinations feature prominently in the news almost daily.

'An honest mistake'

In a statement, Aarons said he had used ChatGPT to summarise trial proceedings after agreeing to take the case last year, but did not grasp the extent to which the chatbot could invent facts. He expressed remorse and hoped the disciplinary board would treat the matter as an honest mistake, describing it as a lesson for professionals who depend on powerful but sometimes unstable technology.

The court also said Aarons had shown a lack of remorse and a lack of concern for his client, and referred him to an attorney disciplinary board for further investigation. The appeal has since been reassigned to Kim Chavez Cook, a New Mexico public defender.

The case adds to a growing pattern across United States courts, where judges at both state and federal levels have disciplined lawyers for submitting AI-generated documents without proper checks. Previous sanctions have typically involved fake case citations and incorrect legal references; the New Mexico filing went further by including fabricated witness testimony in a criminal appeal. Some judges have themselves faced scrutiny over their own use of AI tools.