The brief looked like any other legal filing. It cited witnesses, quoted police testimony, and built an argument. There was just one problem: none of it was real. The witnesses didn't exist. The testimony was invented. And the lawyer who signed it, Stephen Aarons, had never checked.
On Wednesday, the New Mexico Supreme Court held Aarons in direct contempt of court, finding that he submitted a brief containing "false testimony from wholly fabricated witnesses" generated by ChatGPT. The court's order was blunt: Aarons had "demonstrated a lack of remorse and a lack of concern for his client."
What the Court Says Aarons Actually Did
According to the Supreme Court's order, Aarons admitted he did not verify the factual claims or legal authority in his AI-generated brief before signing it and filing it with the court. He also admitted he never informed his client that the brief contained multiple factual and legal misrepresentations.
The brief included fake police testimony and other errors, according to the court. It was filed in a criminal case, meaning a defendant's liberty was potentially affected by arguments built on fiction.
Why This Case Hits Harder Than a Simple Mistake
Legal briefs are sworn documents. When a lawyer signs one, they are certifying that the facts and law cited are accurate to the best of their knowledge. That signature is a promise to the court and to the client.
Here, the court found that promise was broken — not by a small oversight, but by a failure to check anything at all. The fabricated witnesses and fake police testimony weren't minor citation errors. They were the substance of the argument.
For a criminal defendant, that matters enormously. A defense built on invented evidence isn't just weak. It can actively harm the client's case and undermine trust in the entire proceeding.
How AI-Generated Legal Filings Became a Growing Problem
This isn't the first time a lawyer has been caught submitting AI-generated content with fabricated citations. Courts across the United States have increasingly confronted filings that cite cases that don't exist, quote opinions that were never written, and reference witnesses who were never called.
The pattern is consistent: a lawyer uses a generative AI tool, trusts the output without verification, and files it. The AI produces confident, fluent text that looks authoritative — until someone checks.
What makes the New Mexico case stand out is the severity of the response. Direct contempt is a serious finding. It signals that the court viewed the conduct not as negligence, but as an affront to the judicial process itself.
The Client Who Was Never Told
One of the most troubling details in the court's order is that Aarons did not inform his client about the fabricated content. The client — a criminal defendant — had a right to know that the brief filed on their behalf contained false testimony and misrepresentations.
The court's language about Aarons showing "a lack of concern for his client" reflects that breach. A lawyer's duty of loyalty and communication is fundamental. When that duty is ignored, the client is left exposed and unaware.
What the Disciplinary Board Could Do Next
The Supreme Court referred Aarons to a disciplinary board for further proceedings. That board will review the conduct and determine what sanctions, if any, are appropriate.
Possible outcomes range from a reprimand to suspension of his law license or, in severe cases, disbarment. The board's decision will depend on factors including the harm caused, any prior disciplinary history, and whether Aarons shows remorse or takes corrective action.
The court's finding that he lacked remorse may weigh heavily in that process.
Confirmed Facts vs. What Remains Unclear
Confirmed: The New Mexico Supreme Court held Aarons in direct contempt. The brief contained fabricated witness testimony and false police testimony. Aarons admitted he did not verify the content before filing and did not inform his client. He was referred to a disciplinary board.
Unclear: The specific sanctions the disciplinary board may impose. Whether Aarons has retained counsel or issued any public statement. The exact nature of the underlying criminal case and how the fabricated brief affected its outcome.
The Bigger Question Courts Are Now Facing
Judges are increasingly being asked to draw lines around AI use in legal practice. Is using ChatGPT to draft a brief inherently improper? Not necessarily. But filing unverified AI output as fact — and signing your name to it — crosses a clear line.
The legal profession is still adapting. Some courts have issued standing orders requiring lawyers to disclose AI use. Others have imposed sanctions. The New Mexico case adds to a growing body of precedent: the tool may be new, but the duty to verify is not.
What This Means for Lawyers Using AI Right Now
For attorneys, the message is straightforward. AI can be a drafting aid, a research shortcut, or a way to brainstorm. But it cannot replace the lawyer's own verification. Every citation, every fact, every witness reference must be checked against the actual record.
Signing a brief without reading it carefully — or without confirming that its contents are real — is not just risky. In New Mexico, it's now grounds for contempt.
What Happens Next
The disciplinary board will now review the referral. Aarons could face a range of outcomes, from a private reprimand to the loss of his license to practice law. The timeline for that process is not yet clear.
Meanwhile, the case will likely be cited by other courts grappling with similar situations. It sends a signal that judges are willing to treat AI-generated fabrications as serious misconduct, not technical glitches.
Our Take
This case isn't really about ChatGPT. It's about what happens when a lawyer stops doing the one thing that defines the job: checking the facts. AI didn't fabricate witnesses on its own. It produced text that a human being signed and filed without reading closely enough to notice they weren't real.
The court's contempt finding is a reminder that technology changes the tools, not the responsibilities. A lawyer's signature still means something. And when it's attached to fiction, the consequences are real.
Frequently Asked Questions
What did the lawyer actually do wrong?
Stephen Aarons filed a legal brief generated by ChatGPT that contained false testimony from witnesses who did not exist, including fake police testimony. He admitted he did not verify the content before signing and filing it, and he did not tell his client about the misrepresentations.
What does direct contempt of court mean?
Direct contempt is a court's finding that someone has disobeyed or disrespected the court's authority in a way that interferes with the administration of justice. It can lead to sanctions including fines, jail time, or referral to a disciplinary board, depending on the jurisdiction and the conduct.
Can lawyers still use ChatGPT for legal work?
Yes, but with strict verification. AI can assist with drafting, research, and organization, but every fact, citation, and reference must be independently checked against reliable sources. Filing unverified AI output is unethical and can lead to sanctions.
What happens to Aarons now?
The New Mexico Supreme Court referred him to a disciplinary board. That board will investigate and decide whether to impose sanctions, which could include a reprimand, suspension, or disbarment. The process may take months.
