Blab AI
New Mexico Supreme Court Fines Attorney for Filing AI-Generated Fabricated Testimony
The New Mexico Supreme Court imposed a $5,000 fine and a contempt order on criminal‑defense lawyer Stephen Aarons after it discovered that an appellate brief he filed contained testimony from entirely fabricated witnesses. The brief, which the court said was prepared with the assistance of OpenAI’s ChatGPT, included false statements such as a witness claiming the shooter wore dark pants and a white shirt.
Aarons, based in Santa Fe, was representing Oscar Renee Sandoval—convicted of murdering his mother and sentenced to life in prison—in an appeal that the court assigned to public‑defender Kim Chavez Cook on September 2. During a hearing on August 21, Aarons explained that he had fed a computer‑generated transcript and other case materials into ChatGPT, hoping the model would produce a “bullet‑proof summary.” The court found that the brief contained fabricated content and that Aarons had not verified its accuracy.
Justices described the filing as a “monstrosity” and expressed disbelief that the attorney was unaware of the model’s tendency to hallucinate facts. Justice Shannon Bacon asked, “Counsel, do you watch the news? Do you listen to the radio? Do you read anything about what’s going on in the world?” The ruling noted that Aarons showed a lack of remorse and concern for his client.
Aarons issued a statement saying he was remorseful and hoped the disciplinary board would view the incident as an honest mistake. He said the experience was a lesson for professionals who rely on AI tools.
The court will refer Aarons to the attorney disciplinary board for investigation; no additional sanctions were announced at this time. The case follows a growing trend of judicial sanctions against lawyers who submit court documents that contain AI‑generated errors. In recent months, state and federal judges have disciplined attorneys for filing briefs that include fabricated case citations or misquoted statutes.
The New Mexico Supreme Court had previously ordered Aarons to explain how the fabricated material appeared in his primary brief. The decision to fine and hold him in contempt underscores the seriousness with which courts view the use of unverified AI output in legal filings.
OpenAI, the developer of ChatGPT, has not responded to requests for comment. ChatGPT, released in November 2022, is known to produce plausible‑sounding but sometimes incorrect or fabricated information—a phenomenon referred to as hallucination. The model’s use in high‑stakes contexts such as legal research raises concerns about accuracy, confidentiality, and professional responsibility.
This incident highlights the need for attorneys to verify AI‑generated content before submitting it to court and illustrates the broader regulatory and ethical challenges that arise when generative AI tools are used in the legal profession. The appeal of Sandoval remains pending; the court’s ruling does not affect the outcome of the criminal case but may influence Aarons’ ability to represent his client.
As the legal community continues to grapple with AI integration, this case serves as a reminder that the responsibility for accuracy ultimately rests with the human professional. The New Mexico Supreme Court’s decision is part of a broader pattern of judicial oversight aimed at ensuring that AI tools are used responsibly in the courtroom.
Aarons, based in Santa Fe, was representing Oscar Renee Sandoval—convicted of murdering his mother and sentenced to life in prison—in an appeal that the court assigned to public‑defender Kim Chavez Cook on September 2. During a hearing on August 21, Aarons explained that he had fed a computer‑generated transcript and other case materials into ChatGPT, hoping the model would produce a “bullet‑proof summary.” The court found that the brief contained fabricated content and that Aarons had not verified its accuracy.
Justices described the filing as a “monstrosity” and expressed disbelief that the attorney was unaware of the model’s tendency to hallucinate facts. Justice Shannon Bacon asked, “Counsel, do you watch the news? Do you listen to the radio? Do you read anything about what’s going on in the world?” The ruling noted that Aarons showed a lack of remorse and concern for his client.
Aarons issued a statement saying he was remorseful and hoped the disciplinary board would view the incident as an honest mistake. He said the experience was a lesson for professionals who rely on AI tools.
The court will refer Aarons to the attorney disciplinary board for investigation; no additional sanctions were announced at this time. The case follows a growing trend of judicial sanctions against lawyers who submit court documents that contain AI‑generated errors. In recent months, state and federal judges have disciplined attorneys for filing briefs that include fabricated case citations or misquoted statutes.
The New Mexico Supreme Court had previously ordered Aarons to explain how the fabricated material appeared in his primary brief. The decision to fine and hold him in contempt underscores the seriousness with which courts view the use of unverified AI output in legal filings.
OpenAI, the developer of ChatGPT, has not responded to requests for comment. ChatGPT, released in November 2022, is known to produce plausible‑sounding but sometimes incorrect or fabricated information—a phenomenon referred to as hallucination. The model’s use in high‑stakes contexts such as legal research raises concerns about accuracy, confidentiality, and professional responsibility.
This incident highlights the need for attorneys to verify AI‑generated content before submitting it to court and illustrates the broader regulatory and ethical challenges that arise when generative AI tools are used in the legal profession. The appeal of Sandoval remains pending; the court’s ruling does not affect the outcome of the criminal case but may influence Aarons’ ability to represent his client.
As the legal community continues to grapple with AI integration, this case serves as a reminder that the responsibility for accuracy ultimately rests with the human professional. The New Mexico Supreme Court’s decision is part of a broader pattern of judicial oversight aimed at ensuring that AI tools are used responsibly in the courtroom.
