Supreme Court penalizes lawyer for using ChatGPT in murder trial

The New Mexico Supreme Court has found Santa Fe attorney Stephen Aarons guilty of contempt of court. The veteran lawyer, in an appeal for a client convicted of murder, submitted a formal document with content generated by OpenAI’s ChatGPT tool. The lawyer used the popular chatbot to attempt to abbreviate lengthy court transcripts, then submitted the resulting summary to the court without any prior review or reading.

The document contained statements from non-existent people and fabricated claims incorrectly attributed to a real witness. The court’s response was extremely harsh. Judge C. Shannon Bacon publicly asked the lawyer if he even follows the news, as fabricated information or “hallucinations” of artificial intelligence in legal documents have repeatedly made headlines. She added that the lawyer either consciously ignored the phenomenon or recklessly played with the fate of his client. Aarons, who has been in the legal profession for more than four decades, admitted his mistake and said he had mistakenly assumed that the software would produce a reliable summary.

In addition to the public warning, the court ordered the lawyer to pay a fine of 5,000 euros, which will be allocated to the client protection fund, and also referred the case to the body for disciplinary supervision of lawyers. The case testifies to how risky blind trust in new technologies is in the legal sector. At the same time, serious doubts arise about the professionalism of individuals who submit key documents without verification, as such lapses directly threaten the rights of the defendants. This is not an isolated case, as several lawyers have recently faced sanctions for submitting unverified artificial intelligence materials in courts.