A New Mexico defense lawyer has learned a harsh lesson about the limits of artificial intelligence in the legal system. The state Supreme Court held attorney Stephen Aarons in direct contempt of court after he submitted an appeal brief containing false testimony from entirely fabricated witnesses.
Aarons admitted to feeding a trial transcript and other case documents into OpenAI’s o3 model via ChatGPT. He failed to check the output before signing and filing the brief, assuming the AI generated a bulletproof summary.
Fabricated Evidence and AI Hallucinations
The resulting legal document included testimony from officers and witnesses who were never called at trial. It also featured completely made-up accounts regarding threats and the appearance of the shooter in a murder conviction appeal.
State Supreme Court justices heavily criticized Aarons during a disciplinary hearing. Justice C. Shannon Bacon expressed disbelief that an attorney could claim ignorance of AI hallucinations, noting that the phenomenon is widely reported daily.
Legal Responsibility and Court Penalties
The court emphasized that lawyers bear full responsibility for the accuracy of their filings. Whether an error comes from a human assistant or an artificial intelligence tool, the attorney signing the document is legally and ethically bound to verify every fact.
A
arons was barred from appearing before the New Mexico Supreme Court pending disciplinary board proceedings. He also received a $5,000 fine, and the court appointed a public defender to handle a completely new brief for the appellant.
This case serves as a stark warning for professionals across all industries. While large language models offer immense productivity benefits, blind trust in automated output can destroy professional reputations and jeopardize client outcomes.
Key Takeaways
- Direct Contempt: Attorney Stephen Aarons was held in direct contempt by the New Mexico Supreme Court after submitting an appeal brief containing fabricated witness accounts.
- Unverified AI Summaries: Aarons used OpenAI’s o3 model via ChatGPT to summarize transcripts but failed to fact-check the generated document before filing.
- Strict Sanctions: Aarons received a $5,000 fine, was barred from appearing before the state Supreme Court pending disciplinary proceedings, and lost his role on the appeal.
- Lawyer Responsibility: Attorneys are ethically and legally bound to verify every detail in filings, regardless of whether mistakes stem from AI tools or human assistants.
Join our community by subscribing to our Weekly Newsletter to stay updated on the latest AI updates and technologies, including the tips and how-to guides.
(Also, follow us on Instagram (@tid_technology) for more updates in your feed and our WhatsApp Channel to get daily news straight to your Messaging App).
