AGENTRY.NEWSWhat AI Agents Do, Documented.September 26, 2026

Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. AI policy.

New Mexico court fines lawyer $5K for ChatGPT-fabricated witnesses

By
Agentry Newsroom
Published

Stephen D. Aarons, a defense lawyer in New Mexico, submitted an appellate brief containing false testimony from wholly fabricated witnesses in a murder case, and the state's highest court attributed the fabrications to OpenAI's ChatGPT Reuters.

On September 9, 2026, the New Mexico Supreme Court issued a four-page order in Santa Fe finding that Aarons' brief in the appeal of convicted murderer Oscar Renee Sandoval contained made-up police testimony and invented witnesses that did not exist. The fabricated names included Officer Michelle Amarillo, Officer Sanchez, Manal Al-Jibury, Teresa Marquez, Danny Stanton, Linda Stanton, and Mariah Chavez Reuters.

Court Sanctions and Contempt Findings

The court found that Aarons "misrepresented legal authority" in the brief and held him in contempt. The judge ordered Aarons to pay a $5,000 fine Reuters. Beyond the financial penalty, the court referred Aarons for further attorney-discipline review by the state's professional conduct regulator.

In its ruling, the court stated that Aarons "demonstrated a lack of remorse and a lack of concern for his client." The decision underscores the court's view that the violations were not merely technical errors but reflected a failure in professional judgment and accountability to both the judicial system and his represented defendant.

AI Agent Misuse in Legal System

This case represents a concrete instance of an AI agent—ChatGPT—being used to commit fraud within formal legal proceedings. Rather than using the language model as a research or drafting aid, Aarons relied on its output for factual claims about witnesses and police conduct without verification, then submitted those fabrications to the court under penalty of perjury.

The incident was reported by Reuters on September 11, 2026, following the court's order Reuters. Aarons blamed ChatGPT for the errors, though the court's finding focused on his failure to verify the information before filing.

The case demonstrates a critical vulnerability in the adoption of large language models within professions where accuracy and truthfulness carry legal weight. The fabricated witnesses and false police testimony could have influenced appellate review of a murder conviction had they not been detected.

Del dette opslag: