The New Mexico Supreme Court found defense lawyer Stephen Aarons in direct contempt of court for filing a brief that contained AI-fabricated witness testimony, fined him $5,000, barred him from appearing before the court, and ordered a new lawyer for his client.
Short answer: The New Mexico Supreme Court found defense lawyer Stephen Aarons in direct contempt of court for filing a brief that contained AI-fabricated witness testimony, fined him $5,000, barred him from appearing before the court, and ordered a new lawyer for his client.
The New Mexico Supreme Court found a defense lawyer in direct contempt of court after he filed a brief that contained false testimony made up by an AI tool. The lawyer, Stephen Aarons, had been hired by family members to appeal a murder conviction for his client Oscar Renee Sandoval, who was sentenced to life in prison in February 2025. Aarons, who has practiced criminal defense in the state for more than four decades, submitted the brief in August 2025. Weeks later the state moved to strike parts of the filing because it included statements that never occurred in the trial.
Aarons told the court that he had obtained a computer-generated transcript of the trial from a service called Rev.com and then fed that transcript, along with other case documents, into ChatGPT. He said he assumed the model would produce a reliable summary because of its widespread use in fields like law and medicine. He admitted that he did not check the factual claims or legal citations in the AI-generated brief before signing it and filing it, and he never informed his client that the document contained multiple misrepresentations.
The brief contained a series of invented quotes attributed to people who never testified. Aarons acknowledged that it falsely quoted Officer Michelle Amarillo, Officer Sanchez, Manal Al-Jibury and Teresa Marquez as witnesses. It also fabricated statements from Danny Stanton about receiving threats, from Linda Stanton about threats to her husband, and from Mariah Chavez and Teresa Marquez regarding the shooter’s clothing and appearance. In addition to the false testimony, the brief misrepresented several real legal precedents, citing them incorrectly.
The Supreme Court responded by holding Aarons in direct contempt, referring him to the state disciplinary board for further proceedings and ordering him to pay a five-thousand-dollar fine to the State Bar of New Mexico Client Protection Fund. The court barred him from appearing before the Supreme Court while the disciplinary process unfolds. It also directed the public defender’s office to appoint a new lawyer for Sandoval, struck all previous briefs from the record and said the case will be heard during the 2026-27 term.
During the hearing Justice C. Shannon Bacon was especially critical, saying that Aarons had violated at least eight or nine provisions of the lawyer’s code of conduct. She expressed disbelief that he claimed ignorance of AI hallucinations, noting that even her teenage nephew and elderly stepmother are aware of the problem. Bacon accused him of either burying his head in the sand or taking a reckless gamble, both of which are incompatible with the duty of candor to the court. Aarons responded that he had filed the brief a year ago and that much has emerged since then, but the judge pointed out that concerns about AI-generated false citations have been reported regularly for over three years.
In a statement to Ars Technica, Aarons said he regretted the mistake, welcomed the court’s decision to discard the faulty pleading and hoped the disciplinary board would view the error as an honest slip. He described the episode as a lesson for any professional who relies on powerful but imperfect AI tools.
The case underscores a growing risk for anyone who uses generative AI in professional work: the technology can produce convincing yet entirely fabricated information. Lawyers, like other experts, must verify every fact and citation before submitting documents to a court or client, regardless of the source. The episode serves as a reminder that reliance on AI does not relieve the user of the responsibility to ensure accuracy and honesty.
He filed a brief containing AI-generated false testimony and misquoted witnesses, never verified the AI output, and submitted it to the court, violating his duty of candor and court rules.
He used ChatGPT, feeding it a trial transcript from Rev.com and other case documents, and assumed the model would produce a reliable summary because of its widespread use in law and medicine.
The brief invented quotes from Officer Michelle Amarillo, Officer Sanchez, Manal Al-Jibury, Teresa Marquez, Danny Stanton about threats, Linda Stanton about threats to her husband, and Mariah Chavez and Teresa Marquez regarding the shooter’s clothing and appearance.
The court held him in direct contempt, referred him to the state disciplinary board, ordered a $5,000 fine to the State Bar of New Mexico Client Protection Fund, barred him from appearing before the Supreme Court during the disciplinary process, and directed the public defender’s office to appoint a new lawyer for Sandoval.
Source: Ars Technica
New Mexico’s Supreme Court fined defense attorney Stephen Aarons $5,000 and held him in contempt after his appellate brief contained fabricated witness statements and false details about the shooter’s appearance that were generated by ChatGPT.
New Mexico’s Supreme Court fined attorney Stephen Aarons $5,000 and held him in contempt after he submitted a murder-appeal brief containing AI-generated false witnesses and inaccurate police testimony, having failed to verify the fabricated content before filing.
Perplexity began using GPT-6 Astra on September 14, 2026 to draft internal and external messages, tweak software components, and monitor production system health, with human operators now checking in far less often than with earlier models.
NO COMMENTS YET
Comments are open. Have a thought or a question? Share it below.