By Reynand Wu 
By Staff Legal Reporter
September 11, 2026
Main Facts
The New Mexico Supreme Court has penalized a lawyer for submitting an appellate brief that relied heavily on unverified artificial intelligence outputs, resulting in a series of fabricated testimonies and non-existent legal records. In a disciplinary filing released on Wednesday, the state’s high court held attorney Stephen Aarons in contempt of court and issued a $5,000 fine.
The sanctions stem from an appeal Aarons filed on behalf of a client convicted of murder. Rather than conducting exhaustive manual reviews of trial transcripts, Aarons turned to OpenAI’s ChatGPT, expecting the generative tool to produce a "bulletproof summary" of the lower court proceedings. Instead, the AI manufactured a fictional reality: the brief contained entirely fabricated witness accounts, imaginary police testimonies, and false descriptions regarding the shooter’s physical appearance and clothing.
During an August oral argument hearing, Justice C. Shannon Bacon sharply rebuked Aarons for his apparent obliviousness regarding the well-documented phenomenon of AI "hallucinations."
"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 Bacon asked from the bench, according to reports from Reuters. "Because the problem with lawyers relying on AI hallucinations is an above-the-fold story every single day."
The New Mexico Supreme Court’s ruling underscores a growing, systemic crisis within the legal profession: as generative AI tools become ubiquitous, a subset of legal practitioners continues to deploy them without understanding their structural limitations, directly violating rules of professional conduct and duties of competence.
Chronology of Events
To understand how a routine murder appeal devolved into an AI ethics debacle, it is necessary to trace the timeline of the case, the discovery of the fraudulent brief, and the subsequent judicial reckoning.
Phase 1: The Trial and the Appeal
The underlying case involved a violent murder conviction in New Mexico. Seeking to overturn the verdict, attorney Stephen Aarons took on the client’s appeal. In standard appellate practice, an attorney must painstakingly comb through trial transcripts, exhibits, and evidence logs to construct arguments based firmly on the administrative record.
Phase 2: The Reliance on Generative AI
Faced with the dense volume of trial data, Aarons turned to ChatGPT. Operating under the assumption that the large language model could accurately synthesize weeks of courtroom testimony, he prompted the AI to generate summaries and key takeaways to bolster the appeal. However, because generative AI models function by predicting statistically probable strings of text rather than retrieving verified factual databases, ChatGPT filled gaps in its "memory" with completely fabricated details.
Phase 3: The Filing of the Brief
Without cross-referencing the AI-generated output against the actual, physical transcripts of the trial, Aarons incorporated the summaries directly into his formal appellate filings submitted to the New Mexico Supreme Court. The brief asserted "facts" that had never occurred in a court of law, including detailed statements from witnesses who did not exist and fabricated recollections from police officers.
Phase 4: The Discovery and August Hearing
The discrepancies did not go unnoticed. Opposing counsel and the justices reviewing the brief identified glaring inconsistencies between the cited material and the actual trial record. During the August hearing, Justice Bacon confronted Aarons about the provenance of the brief. Cornered, Aarons admitted to utilizing ChatGPT in an effort to streamline his workflow.
Phase 5: The Official Sanctions (September 11, 2026)
Following deliberations, the New Mexico Supreme Court issued its formal ruling on Wednesday. The court found Aarons in direct contempt, slapped him with a $5,000 fine, and rebuked him for failing to fulfill his foundational duty to verify the factual claims and legal authority underpinning court submissions.
Supporting Data & Industry Context
The New Mexico incident is far from an isolated anomaly. It is part of a mounting wave of technological malpractice cases sweeping courts across the United States as generative AI adoption outpaces institutional training and regulatory frameworks.
- Rising AI Adoption vs. Competence: According to the American Bar Association’s (ABA) 2026 Legal Industry Reports, an increasing percentage of law firms and solo practitioners are integrating generative AI into their daily operations to reduce overhead and accelerate document drafting.
- The "Hallucination" Trap: Generative AI models are fundamentally linguistic probability engines. When asked for obscure legal precedents, specific citations, or granular factual summaries, they frequently hallucinate—inventing convincing-sounding case names, fictitious judges, and nonexistent statutory codes.
- Precedent Cases Across Jurisdictions:
- The New York Federal Court Incident (Prior Years): Two law firms faced severe judicial backlash after submitting a brief filled with bogus legal citations and quotations entirely generated by ChatGPT, leading to heavy financial penalties.
- The MyPillow Legal Team Sanctions: Attorneys representing Mike Lindell and MyPillow were similarly sanctioned and fined by courts for incorporating AI-generated misquotes and fictitious legal authorities into their filings.
- The New Mexico Ruling: The New Mexico Supreme Court’s decision to elevate the infraction to contempt and levy a direct financial penalty signals that courts are moving past warnings and adopting zero-tolerance policies for unverified AI submissions.
Official Responses and Statements
Legal professionals, regulatory bodies, and the sanctioned attorney have weighed in on the fallout from the New Mexico high court’s decision.
Attorney Stephen Aarons
In a statement provided to Reuters following the release of the court’s order, Aarons expressed remorse while attempting to contextualize the error as an oversight rather than an act of bad faith.
"I am remorseful but hopeful that the disciplinary board takes into account it was an honest mistake," Aarons stated.
Despite his pleas that the incident was merely a lapse in judgment, legal ethicists note that an attorney’s ethical duty of competence under Model Rule 1.1 requires staying abreast of the risks and benefits of relevant technology. Claiming ignorance of AI hallucinations no longer serves as an acceptable defense in modern American jurisprudence.
Judicial Perspective
Through their actions and on-the-record admonishments—epitomized by Justice Bacon’s pointed query regarding whether lawyers consume basic news or media—courts are signaling that basic digital literacy is now a baseline requirement for practicing law. Judges across federal and state systems are increasingly treating the uncritical submission of AI-generated text as a direct affront to the integrity of the judicial process.
Implications for the Legal Profession
The New Mexico Supreme Court’s ruling against Stephen Aarons carries profound implications for the future of legal practice, technology regulation, and professional accountability.
1. The Redefinition of Legal Competence
Traditionally, legal competence was measured by an attorney’s understanding of statutory interpretation, case law, and courtroom procedure. In the era of generative AI, competence explicitly encompasses technical due diligence. Lawyers can no longer blindly trust software tools to perform substantive legal or factual research without human verification. Every citation, witness statement, and factual assertion must be independently audited against primary sources.
2. Mandatory Standing Orders and Local Rules
An increasing number of federal judges, magistrate judges, and state courts have begun issuing standing orders explicitly requiring lawyers to certify whether generative AI tools were used in drafting court documents, and if so, whether a human attorney verified every word. Failure to adhere to these emerging administrative requirements risks immediate sanctions, contempt citations, and potential referral to state bar disciplinary committees.
3. Reputational and Career Damage
Beyond financial fines and contempt charges, incidents involving AI hallucinations inflict catastrophic damage on an attorney’s professional reputation. In a digital age where court filings are instantly searchable public records, being publicly named in a supreme court order for submitting fabricated witness testimony can permanently impair client trust and professional standing.
4. The Broader Technological Paradox
The legal industry faces a profound paradox: while clients demand greater efficiency and lower billing hours—incentivizing the adoption of AI automation—courts demand absolute accuracy and meticulous human oversight. Navigating this tension requires law firms to invest heavily in specialized, legal-specific AI platforms grounded in verified databases, rather than relying on consumer-grade models like ChatGPT, which are notoriously prone to invention.
Ultimately, the New Mexico Supreme Court’s reprimand serves as a stark warning to the legal community: artificial intelligence can assist in the practice of law, but it can never replace the human intellect, skepticism, and ethical responsibility required to uphold the integrity of the justice system.
Siti Muinah 
Asro 
Suro Senen 