New Mexico Lawyer Fined for ChatGPT-Fabricated Court Filing
The Supreme Court sanctioned an attorney for submitting an appeal containing fake witnesses and false testimony.
Updated on Sept. 25, 2026 in Artificial Intelligence
Live Poll
Do you trust the accuracy of legal documents that are partially written by artificial intelligence?
On September 9, 2026, the New Mexico Supreme Court fined lawyer Stephen D. Aarons $5,000 for submitting legal documents generated by ChatGPT that contained four fabricated witnesses. The court subsequently struck the briefs and appointed the New Mexico Public Defender's Office to represent his client, Oscar Renee Sandoval.
Why it matters
The case highlights the risks of using generative AI for legal research without human verification, a practice that risks compromising judicial integrity. It reinforces the necessity of compliance with professional standards, such as the American Bar Association's 2024 guidance on AI-generated work.
The filing included four fabricated witnesses and misrepresented two court decisions. Stephen D. Aarons admitted he failed to perform any verification of the factual claims or legal references generated by the model.
The players
Stephen D. Aarons
The lawyer responsible for the submission of fabricated legal documents in a murder conviction appeal.
New Mexico Supreme Court
The highest judicial body in the state responsible for oversight of legal practitioners.
Oscar Renee Sandoval
The defendant in a murder conviction appeal case who required a change in representation.
New Mexico Public Defender's Office
The state agency appointed to assume representation of the defendant following the court ruling.
American Bar Association
A national professional association that provides guidance and ethical frameworks for legal practitioners.
The details
The attorney utilized ChatGPT to generate arguments for a murder conviction appeal, but failed to cross-reference the output against primary sources. Large language models — deep-learning systems that predict the next word in a sequence based on vast training datasets — can exhibit 'hallucinations,' where the system generates convincing but entirely fictional information. The Supreme Court of New Mexico found that these unchecked submissions violated the duties of competence and candor required of officers of the court.
Timeline
February 2025: Oscar Renee Sandoval was sentenced to life in prison.
2024: The American Bar Association issued Formal Opinion 512 regarding lawyer use of AI.
September 9, 2026: The Supreme Court issued the disciplinary order regarding the filings.
The Tech Race
This case follows the precedent set by the American Bar Association's 2024 Formal Opinion 512, which mandates strict human oversight of AI-generated legal work. It marks a departure from earlier legal practices by explicitly enforcing these standards to penalize the misuse of generative models.
This ruling affects the standard of evidence required in legal filings by reinforcing that AI-generated citations must be verified against official databases. For practitioners, it confirms that ignorance of AI limitations is not a valid defense for procedural negligence.
The takeaway
Legal professionals must treat output from generative models as unverified drafts that require manual verification against primary sources. Future developments in this area will be shaped by the Disciplinary Board's forthcoming review of the attorney's professional license.
What happens next
The Disciplinary Board is scheduled to conduct a further review of Stephen D. Aarons' conduct, which will determine if additional professional sanctions are necessary beyond the current fine and court suspension.
Further reading
For broader context on the ethical guidelines governing automation in law, visit the Artificial Intelligence section.
Source note: This article includes information reported by Las Cruces Sun-News.
Live Poll
Do you trust the accuracy of legal documents that are partially written by artificial intelligence?
