US Lawyer Sanctioned Over ChatGPT-Fabricated Witnesses
New Mexico's Supreme Court fined attorney Stephen Aarons $5,000 and held him in contempt of court over a filing containing fabricated witness testimony.
Aarons had submitted an appeal brief for a client convicted of murder that cited statements from several witnesses, including supposed police officers, who the court says never existed.
According to Aarons, he fed a computer-generated trial transcript into ChatGPT expecting a reliable summary, and did not verify the output before filing it.
Justice C. Shannon Bacon strongly criticized him, asking whether he had missed the frequent news coverage of AI errors in legal filings.
Aarons said he had not understood how badly AI systems can fabricate facts and described the incident as an honest mistake.
The court struck all previous filings from the case, appointed a new public defender for the client, and referred Aarons to a disciplinary board for lawyers.
Similar cases involving AI-fabricated legal documents have been increasing in US courts, following earlier sanctions against other law firms.
WHY IT MATTERS
The case shows that AI systems can produce convincing but entirely false information even in serious, professional settings.
WHAT YOU CAN DO NOW
- Always double-check AI answers Compare key names, numbers, and claims from an AI response against an independent, trustworthy source. Never rely on AI output alone.
- Treat AI as a helper, not a source Think of AI-written text as a first draft that you must review yourself. For legal, medical, or financial matters, AI cannot replace an expert.
- Get important documents checked by experts Have contracts, applications, or other important papers reviewed by a real professional. This helps catch anything AI may have invented.
- Stay cautious with polished summaries A smooth, complete-sounding AI summary is not proof of accuracy. The more polished it looks, the more worth it is to double-check the details.