US Lawyer Fined $5,000 for Submitting ChatGPT-Generated Fabricated Testimony to Court
The New Mexico Supreme Court fined defense attorney Stephen Ahrens $5,000 for contempt of court after he submitted a murder appeal brief containing fabricated witnesses and fake police testimony generated by ChatGPT. Ahrens admitted he relied on ChatGPT to summarize trial transcripts without verifying the accuracy of the generated content. This case highlights the escalation of risks associated with unverified generative AI usage in legal settings, moving from fabricated legal citations to invented factual evidence. It underscores the strict professional duty of attorneys to review AI output as judiciary bodies increasingly penalize AI-driven misconduct. Beyond the $5,000 fine, the Supreme Court referred Ahrens to the disciplinary board for further investigation, citing a lack of remorse and care for his client. During the hearing, judges expressed disbelief that Ahrens was unaware of widely publicized LLM hallucination risks when he fed trial transcripts into ChatGPT expecting an error-free summary.
## BACKGROUND
Generative AI tools based on Large Language Models (LLMs), such as ChatGPT, are prone to AI hallucinations—instances where the model confidently generates plausible-sounding but completely false information. In legal practice, ethical guidelines mandate that attorneys independently verify all facts, citations, and testimony submitted to the court, as unverified submissions undermine the justice system and carry strict sanctions.