New Mexico fines lawyer $5K for AI‑generated fake testimony
According to The Verge, the New Mexico Supreme Court fined attorney Stephen Aarons $5,000 and held him in contempt after his appeal relied on ChatGPT‑generated witnesses and fabricated police testimony. The ruling underscores how quickly AI‑driven shortcuts can collide with professional ethics in the courtroom.
The brief that sparked the fine
Aarons filed an appeal in a murder conviction case that, according to the court filing, "contained false testimony from wholly fabricated witnesses" and misdescribed the shooter’s clothing and appearance. During an August hearing, Justice C. Shannon Bacon asked, "Counsel, do you watch the news?" after Aarons admitted he used ChatGPT hoping it would produce a "bulletproof summary" of the trial. The justice’s remark highlighted a daily‑news narrative: lawyers are increasingly trusting AI without verifying its output.
How AI hallucinations happen in legal writing
AI‑generated text models like ChatGPT predict the next word based on patterns in massive datasets. When asked for a summary of a case, the model may invent details—so‑called hallucinations—because it tries to fill gaps with plausible‑sounding content. In a legal brief, a hallucinated witness name or a fabricated police quote looks like a real citation until someone cross‑checks it. The model has no built‑in fact‑checking; it simply assembles text that statistically fits the prompt.
Prior incidents and the growing pattern
| Year | Court | Issue | Penalty |
|---|---|---|---|
| 2023 | U.S. District Court (MN) | Two firms submitted briefs with numerous false legal citations | Judge reprimand, no monetary fine reported |
| 2023 | Minnesota Court of Appeals | MyPillow’s Mike Lindell lawyers used AI‑generated misquotes | Fines imposed (amount not disclosed) |
| 2024 | New Mexico Supreme Court | Stephen Aarons used ChatGPT‑fabricated witnesses | $5,000 fine, contempt finding |
These cases show a trend: as more attorneys adopt generative AI, courts are beginning to enforce traditional verification standards more strictly. The penalties vary, but the consistent message is that AI does not replace the lawyer’s duty to confirm every fact.
Why the fine matters for the legal profession
The disciplinary action sends a clear signal to bar associations and law firms: reliance on AI tools without rigorous oversight can constitute professional negligence. Lawyers are bound by rules of competence and diligence; using a model that can hallucinate violates those rules when the output is submitted as fact. The fine also raises cost considerations. While AI can cut drafting time, the risk of costly sanctions may outweigh time savings for high‑stakes matters like criminal appeals.
What actually changes: the hidden trade‑off
The trade‑off nobody spells out is between speed and accountability. In practice, an attorney who spends an hour drafting a brief with ChatGPT may save a few hours of research, but the model’s propensity to fabricate data creates a hidden compliance cost. If a fabricated citation is caught, the lawyer faces not only disciplinary fines but also reputational damage that can affect future client trust. Moreover, law firms may need to invest in additional review layers—human editors, citation‑checking software, or AI‑assistants with built‑in verification—to mitigate the risk. Those extra steps erode the time‑saving benefit that initially attracted lawyers to generative AI.
Who gains and who loses? Early adopters who pair AI with strict fact‑checking can keep the productivity edge. Conservative firms that continue to rely solely on traditional research avoid the immediate risk but may fall behind in efficiency. The courts, meanwhile, gain a clearer enforcement precedent that could shape future bar‑association guidelines.
What to watch next? Bar associations across the U.S. are drafting advisory opinions on AI use. Expect formal opinions from the ABA (American Bar Association) and state bars that define acceptable workflows, such as mandatory citation verification or disclosure of AI assistance in filings. Keep an eye on emerging legal‑tech platforms that integrate real‑time case‑law databases with AI, promising to reduce hallucinations by grounding output in verified sources.
Practical steps for lawyers today
- Treat AI output as a draft, not a final product. Run every citation, witness name, and factual claim through a trusted source—court records, official transcripts, or a reputable database.
- Document the workflow. Keep a log of prompts used, AI responses, and the verification steps you performed. This creates a defensible trail if a court questions the brief.
- Use AI tools built for law. Platforms like Casetext CoCounsel or Westlaw Edge incorporate citation checking that flags unsupported references, reducing hallucination risk.
- Disclose AI assistance where required. Some jurisdictions may soon mandate a note in filings stating that AI was used in drafting.
- Stay informed about bar guidance. Subscribe to updates from your state bar’s ethics committee to adapt quickly to new rules.
By integrating these habits, attorneys can enjoy AI’s efficiency without falling into the trap that caught Aarons.
