If you’re a lawyer leaning on ChatGPT to speed up your briefs, read this before you file anything. A New Mexico attorney just got fined $5,000 and held in contempt because his AI-generated appeal was stuffed with witnesses who never existed. According to The Verge AI, the state’s Supreme Court punished Stephen Aarons for filing a brief in a murder appeal that contained “false testimony from wholly fabricated witnesses,” plus made-up details about the shooter’s clothing and appearance.
The risk here isn’t hypothetical anymore. It’s a court order with your name on it.
What happened
The Verge AI reports that the court’s Wednesday filing faulted Aarons for failing to “verify the factual claims and legal authority in his AI-generated brief.” This wasn’t a bad citation buried in a footnote. The AI fabricated entire witnesses and put fake words in their mouths, then Aarons filed it in an appeal tied to a client’s murder conviction.
At an August hearing, Aarons admitted he used ChatGPT and expected it to produce a “bulletproof summary” of the trial. Justice C. Shannon Bacon wasn’t having it. “Counsel, do you watch the news? Do you listen to the radio? Do you read anything about what’s going on in the world?” she asked, per Reuters. “Because the problem with lawyers relying on AI hallucinations is an above-the-fold story every single day.”
Aarons says he’s “remorseful but hopeful that the disciplinary board takes into account it was an honest mistake.”
Why this keeps happening
What stands out here is how predictable this was. Chatbots don’t retrieve facts from a database. They generate text that looks right, which means they’ll invent a plausible-sounding witness or a citation that reads like real case law but points to nothing. Lawyers call these “hallucinations.” Courts call them grounds for sanctions.
And Aarons is far from alone. As The Verge AI notes, the pattern is now routine:
- Last year, a judge slammed two law firms for a brief with “numerous false, inaccurate, and misleading legal citations and quotations.”
- Lawyers for MyPillow’s Mike Lindell were fined for submitting AI-generated misquotes and fake citations.
- The New Mexico case now adds fabricated murder-trial testimony to the list.
The common thread: the tool wasn’t the problem. Skipping verification was.
Why it matters
This is significant because it marks a shift from embarrassment to consequences. Early AI-citation blunders got lawyers a stern lecture and a viral news cycle. Now judges are handing down fines and contempt findings, and they’re doing it in serious criminal matters where a client’s freedom is on the line.
For the broader industry, it’s a reminder that “AI did it” is not a defense anywhere. The professional on record owns the output. That principle is going to ripple far beyond law, into medicine, finance, journalism, and any field where a confident wrong answer carries real cost.
What to do about it
If you use AI in any high-stakes writing, treat every factual claim it produces as unverified until you check it yourself. A few habits that protect you:
- Verify every citation, quote, and name against the primary source. If you can’t find it, it doesn’t exist.
- Use AI for drafting and structure, not for facts. Let it organize your argument, not invent your evidence.
- Assume confident tone means nothing. Hallucinations read exactly like real information.
- Build a checking step into your workflow before anything gets filed, published, or sent.
The tools aren’t going away, and they’re genuinely useful for the grunt work. But the accountability sits with the human every time. Courts have now made that expensive to forget.
Expect more sanctions like this, and expect them to get steeper as judges lose patience. You can read the full details at the original source.