New Mexico Lawyer Sanctioned for ChatGPT-Fabricated Legal Brief
TECH

New Mexico Lawyer Sanctioned for ChatGPT-Fabricated Legal Brief

27+
Signals

Strategic Overview

  • 01.
    The New Mexico Supreme Court held Santa Fe defense attorney Stephen Aarons in direct contempt and fined him $5,000 in a September 9, 2026 order over a murder-appeal brief containing witnesses fabricated by ChatGPT.
  • 02.
    ChatGPT invented four fabricated witnesses, including two purported police officers, Officer Michelle Amarillo and Officer Sanchez, plus civilian witnesses Manal Al-Jibury and Teresa Marquez, and attributed false statements to them and to real witnesses.
  • 03.
    The court struck all of Aarons' earlier briefs, appointed public defender Kim Chavez Cook as new counsel, barred Aarons from appearing before the New Mexico Supreme Court pending a Disciplinary Board investigation, and ordered him to pay the $5,000 sanction to the State Bar of New Mexico Client Protection Fund within 30 days.
  • 04.
    Aarons said he fed trial transcripts into ChatGPT (using OpenAI's o3 model) expecting a 'bulletproof summary' but did not verify the AI-generated content before filing it, and the court found he had shown a lack of remorse and concern for his client.

Deep Analysis

What Actually Went Wrong Inside the Brief

Aarons, representing client Oscar Sandoval on a murder appeal, fed trial transcripts and case materials into ChatGPT running OpenAI's o3 model, expecting it to produce a 'bulletproof summary' of the proceedings [1]. Instead, the tool invented four witnesses who never testified, including two purported police officers, Officer Michelle Amarillo and Officer Sanchez, alongside two civilian names, Manal Al-Jibury and Teresa Marquez [1]. The fabricated testimony included granular invented details, such as a description of the shooter wearing dark pants and a white shirt [2], and the filing also attributed false statements to real witnesses, blending fact and fabrication in a way that would be nearly impossible to catch without cross-checking the original transcript [1].

A Widening Pattern, Not an Isolated Slip

The case echoes the 2023 Mata v. Avianca episode, in which two attorneys and their firm were fined $5,000 for citing six ChatGPT-invented court cases [7], but it marks an escalation from fabricated case law to fabricated trial testimony in a capital murder case. Courts have hardened their responses since then - a Central District of California special master struck briefs and ordered $31,100 in fees after a filing mixed erroneous outputs from three different AI tools [5], and a Wyoming federal judge revoked an attorney's pro hac vice admission over eight fabricated citations [5]. An industry tracker recorded roughly 200 AI-hallucination sanctions cases in mid-2025, growing to 719 by January 2026, 1,227 by early April 2026, and 1,598 by June 9, 2026 [6].

The Court's Rebuke and What Aarons Loses

Justice C. Shannon Bacon pressed Aarons at the sanctions hearing, asking, 'Counsel, do you watch the news? Do you listen to the radio? Do you read anything about what's going on in the world?' [2]The court held him in direct contempt and fined him $5,000, payable within 30 days to the State Bar of New Mexico Client Protection Fund [1]. It struck all of Aarons' earlier briefs, appointed a public defender, and barred him from appearing before the state's highest court pending a Disciplinary Board investigation [1], finding that he had shown 'a lack of remorse and a lack of concern for his client' [2]. Aarons said he was 'remorseful but hopeful that the disciplinary board takes into account it was an honest mistake' [1].

Collateral Damage: A Murder Appeal Sent Back to Square One

The client, Oscar Renee Sandoval, is serving a life sentence for murdering the mother of his children [3]. His appeal must now be re-briefed from scratch by newly appointed public defender Kim Chavez Cook [4], adding delay to a capital case through no fault of his own, purely because his own lawyer filed unverified AI output with the court.

Historical Context

2023-06
Attorneys Steven A. Schwartz and Peter LoDuca, and their firm Levidow, Levidow & Oberman, were fined $5,000 by Judge P. Kevin Castel after submitting a brief citing six entirely fabricated ChatGPT-generated cases; considered the first major case reshaping how courts and firms think about AI use in legal practice.
2025-05-14
A 10-page supplemental brief contained nine citation errors, including two completely fabricated cases mixing CoCounsel, Westlaw Precision, and Google Gemini outputs; Special Master Michael Wilner struck the briefs, denied discovery relief, and ordered $31,100 in fees.
2025
Attorney Jisuh Lee's factum contained four incorrect citations, including two to non-existent cases; Judge Fred Myers ordered her to show cause on contempt charges.
2025
Three attorneys filed motions citing nine cases, eight of which did not exist; a Wyoming federal judge revoked the drafting attorney's pro hac vice admission and fined him $3,000, with $1,000 fines for the two signing attorneys.
2026-06-09
A running tracker of AI-hallucination sanctions cases showed growth from roughly 200 cases in mid-2025 to 719 by January 2026, 1,227 by early April 2026, and 1,598 by June 9, 2026, with penalties escalating from fines to fee awards, revoked admissions, suspensions, and a canceled trial.
2026-09-09
The New Mexico Supreme Court issued its contempt order and $5,000 fine against Aarons over the ChatGPT-fabricated murder-appeal brief, part of the broader escalating pattern of judicial sanctions for unverified AI-generated legal content.

Power Map

Key Players
Subject

New Mexico Lawyer Sanctioned for ChatGPT-Fabricated Legal Brief

ST

Stephen Aarons

Santa Fe, New Mexico-based private defense attorney sanctioned for filing the ChatGPT-fabricated appellate brief

NE

New Mexico Supreme Court

Issued the September 9, 2026 order holding Aarons in contempt, fining him, striking the briefs, and barring him from the court pending disciplinary review

OS

Oscar Renee Sandoval

Aarons' client, convicted of murdering the mother of his children and sentenced to life in prison; his appeal was disrupted and must be re-briefed

KI

Kim Chavez Cook

New Mexico public defender appointed as new counsel to re-handle Sandoval's appeal after Aarons was removed

DI

Disciplinary Board of the New Mexico Supreme Court

Body to which Aarons was referred for further investigation and potential professional discipline

OP

OpenAI / ChatGPT (o3 model)

AI tool used by Aarons to summarize trial transcripts, which generated the fabricated witnesses and false testimony central to the sanction

Fact Check

7 cited
  1. [1] New Mexico Lawyer Sanctioned Over ChatGPT-Fabricated Witnesses
  2. [2] US lawyer cites fake witnesses in murder case, blames ChatGPT
  3. [3] Court sanctions lawyer over ChatGPT fabricated testimony
  4. [4] New Mexico lawyer held in contempt over ChatGPT-fabricated testimony in murder appeal
  5. [5] AI Hallucinations Strike Again: Two More Cases Where Lawyers Face Judicial Wrath for Fake Citations
  6. [6] AI Legal Hallucination Audit
  7. [7] Mata v. Avianca, Inc.

Source Articles

Top 5

THE SIGNAL.

Analysts

Questioned Aarons's claimed unfamiliarity with widely reported AI hallucination problems in legal filings, asking him directly: "Counsel, do you watch the news? Do you listen to the radio? Do you read anything about what's going on in the world?"

Justice C. Shannon Bacon (New Mexico Supreme Court)
Presiding justice at the sanctions hearing

Warned that AI hallucinations risk contaminating judicial orders themselves if not caught, saying of one near-miss: "That's scary. It almost led to the scarier outcome (from my perspective) of including those bogus materials."

Special Master Michael Wilner (U.S. District Court, Central District of California)
Court-appointed special master in an unrelated 2025 AI-fabrication sanctions matter

Reaffirmed that lawyers bear personal responsibility to verify cited authority regardless of the tool used to find it: "It is the lawyer's duty to read cases before submitting them to a court as precedential authorities."

Hon. Fred Myers (Ontario Superior Court)
Judge overseeing a separate 2025 AI-citation sanctions matter
The Crowd

NEW: The New Mexico Supreme Court held a lawyer in direct contempt after he submitted a ChatGPT-generated brief containing “false testimony from wholly fabricated witnesses,” including invented police testimony and other errors. The state’s highest court referred the lawyer to a...

@@TheInsiderPaper355

A New Mexico defense lawyer used ChatGPT to summarize trial materials for a murder appeal, and it invented fake witnesses he filed without checking. He was fined $5,000 and removed from the case A judge asking "do you not watch the news?" is a rough way to get sanctioned

@@interesting_aIl65

A justice asked, ‘Counsel, do you watch the news?’ when a lawyer admitted to relying on ChatGPT without understanding the risks.

@@verge24
Broadcast
CHATGPT INVENTED TESTIMONY? Can AI Failures Deny Effective Counsel? | Strickland v. Washington

CHATGPT INVENTED TESTIMONY? Can AI Failures Deny Effective Counsel? | Strickland v. Washington

8 Stories: New Mexico Court Punishes Lawyer Over AI Appeal | AI News Today 12 Sep 2026

8 Stories: New Mexico Court Punishes Lawyer Over AI Appeal | AI News Today 12 Sep 2026

New Mexico Fines a ChatGPT Lawyer $5,000 / Amazon Sells Ad Space Inside ChatGPT|2026.09.12 AI News

New Mexico Fines a ChatGPT Lawyer $5,000 / Amazon Sells Ad Space Inside ChatGPT|2026.09.12 AI News

New Mexico Lawyer Sanctioned for ChatGPT-Fabricated Legal Brief — AI News | Agentic Brew