In August 2025, Aarons submitted a brief that included testimony attributed to people who did not exist in the case record. Among them were Officer Michelle Amarillo, Officer Sanchez, Manal Al-Jibury, and Teresa Marquez. The filing also assigned invented statements to several real witnesses, including comments about threats and descriptions of the shooter.
At an August 21 hearing, Aarons told the court that he had uploaded a Rev.com transcript of the trial along with other case materials into ChatGPT, using OpenAI’s o3 model. He said he relied on the system to summarize the proceedings and did not independently check the output before filing the brief.
“I assumed that it generated a bulletproof summary of proceedings,” Aarons told the court.
He also acknowledged that he did not tell Sandoval about his use of ChatGPT. The brief additionally misstated some real legal precedents, although the cited cases themselves were not fabricated.
The court focused less on the technology itself than on Aarons’ responsibility for the filing. Justice C. Shannon Bacon told him there were “at least eight or nine provisions in the code of conduct” that his conduct implicated and rejected his argument that he had not understood how often large language models can invent information.
“So counsel, do you watch the news? Do you listen to the radio? Do you read anything about what's going on in the world?” Bacon said. “Because the problem with lawyers relying on AI hallucinations is an above-the-fold story every single day.”
Chief Justice Julie Vargas similarly pushed back when Aarons attempted to broaden the hearing into a discussion of AI policy, emphasizing instead the consequences for Sandoval, who remains in custody.
Justice Michael Vigil said the choice of tool did not remove the lawyer’s duty to review the work before submitting it.
“It doesn't matter whether you use a C-student lawyer or a A-student lawyer if you didn't check their work before you filed the brief,” Vigil said.
The court concluded that Aarons had “demonstrated a lack of remorse and a lack of concern for his client.” His $5,000 fine is payable to the State Bar Client Protection Fund.
The consequences extend beyond the monetary penalty. Aarons has been removed from the appeal, while the prior briefing has been discarded and must be redone. The case will now continue during the 2026-27 term with new representation from the public defender’s office.
Bacon repeatedly pointed to the impact on Sandoval, whose appeal has been delayed while the defective filing is replaced.
“Your client is the one suffering because of this far more than you will ever suffer,” she told Aarons.
After the ruling, Aarons said he regretted what happened and hoped the disciplinary process would treat the filing as an honest error.
“I am remorseful but hopeful that the disciplinary board takes into account it was an honest mistake,” he said. “It is a lesson learned for all professionals who rely upon this powerful but sometimes unstable technology.”
He also suggested that the court adopt a standing requirement for attorneys to certify their compliance when AI tools are used in legal filings.
The case adds to a growing collection of court sanctions involving lawyers who submitted AI-generated material without verifying it. Many earlier incidents involved nonexistent legal citations or inaccurate descriptions of cases. Aarons’ filing drew particular scrutiny because the errors extended into fabricated witness testimony in a criminal appeal.
For the New Mexico Supreme Court, however, the central issue was not whether ChatGPT produced the false material. It was that a lawyer signed and submitted the brief without confirming that the underlying testimony and record were accurate.
This analysis is based on reporting from The Guardian.
Image courtesy of Dave Cathey.
This article was generated with AI assistance and reviewed for accuracy and quality.
