The New Mexico Supreme Court has found an experienced criminal defense lawyer, Stephen Aarons, in contempt of court for submitting a brief filled with fabricated witness testimonies generated by ChatGPT. The brief included false police testimony and other legal and factual errors, leading to a referral to a disciplinary board for further proceedings.

Aarons admitted to not verifying the information before filing it, and that he did not inform his client about the mistakes. He was hired by the family of Oscar Renee Sandoval, who was sentenced to life in prison for a murder conviction. The lawyer fed a trial transcript into ChatGPT, leading to the submission of several pieces of false testimony, including those from non-existent officers and civilians, and misrepresentations of legal authority.

The case highlights the potential risks of relying on AI for legal research and the importance of human oversight. It also raises concerns about the integrity of legal proceedings when AI-generated content can be so convincing.

The incident could set a precedent for future legal disputes over the use of AI in legal documents, and may prompt stricter guidelines for the use of such tools in court.