Nevada judge offers sanctions-or-education ultimatum after ChatGPT fake citations

AI Model: ChatGPT (OpenAI) Legal TechHallucinationsSanctionsProfessional ResponsibilityU.S. Courts

I. Executive Summary

A Nevada state-court judge found that a filing by two defense lawyers contained at least 14 apparently fictitious case citations, along with other misquoted or misrepresented authorities. After the lawyers and their firm (Cozen O’Connor) acknowledged that ChatGPT had been used in drafting and an uncorrected draft was filed, the judge offered a choice between monetary sanctions (plus potential removal and bar referral) or a remedial “education” route involving written explanations to law school deans and bar officials and volunteering to speak about AI and professional conduct.

II. Key Facts

Judge David Hardy (Washoe County, Nevada) identified ≥14 citations that appeared fictitious.
Cozen O’Connor said an associate used ChatGPT to draft/edit and an early draft was filed by mistake.
The judge offered $2,500 per lawyer (plus removal/referral) or an alternative remedial program (letters + speaking).
Cozen stated it maintains a strict policy restricting public AI tools for client work and requiring attorney verification.

III. Regulatory & Ethical Implications

The matter illustrates an enforcement trend: courts are moving from simple reprimands to structured deterrents (costs, disqualification, bar referrals, and reputational/educational remedies). For compliance programs, it reinforces the need for explicit AI-use policies, mandatory citation verification controls, and auditability of drafting workflows.

IV. Media Coverage & Sources