Neil Sands
The Supreme Court has warned self-represented litigants who submit Artificial Intelligence (AI)-generated “hallucinations” to court that they could face contempt proceedings.
The court, in declining leave to appeal a case that originated in the Family Court, noted that the applicant, John Jones, “cited a number of authorities which appear to have been hallucinated by an AI application”.
“Misuse of AI in legal proceedings has serious implications for the administration of justice and public confidence in the justice system,” the Supreme Court panel, comprising Justices Ellen France, Stephen Kós, and Forrie Miller said.
“Persons filing submissions in court must ensure all authorities referred to are genuine and correctly cited.”
The judges pointed to Courts of New Zealand’s guidelines on AI use for non-lawyers, which warn that AI chatbots “can make up fake cases, citations and quotes, or refer to legislation, articles or legal texts that do not exist” – commonly referred to as hallucinations.
However, the Supreme Court decision went further that the guidelines by pointing out the potential consequences for careless use of AI in court.
“Reliance on false citations, including the unverified outputs of AI applications, may in serious cases amount to obstruction of justice or contempt of court,” it said.
Obstruction of justice carries a maximum prison term of seven years undet the Crimes Act 1961, while higher courts can impose a six-months’ imprisonment or a fine of up to $25,000 on individuals under the Contempt of Court Act 2019.
The Supreme Court’s remarks on AI could be interpreted as the judiciary losing patience with the amount of AI-generated “slop” reaching the courts through self-represented litigants.
A global database of cases in which AI hallucinations have been cited lists six in New Zealand, including the latest instance in the Supreme Court, all involving self-represented litigants.
The official AI usage guidelines for self-represented litigants, lawyers, judges and court support staff do not require proactive disclosure of whether submissions have been prepared using AI.
However, Bell Gully partner Rosemary Wooders told a TLANZ-organised webinar on AI use last week that there was an argument to change the guidelines to help identify potentially problematic submissions.
“The generative AI guidelines produced by the court were done some almost three years ago now, and there is a specific statement in there saying you do not need to disclose use of a Gen AI chatbot as a matter of course, unless asked by the court or tribunal,” she said.
“Now, I do think it would be helpful, potentially, if you were a litigant, to disclose, just so that the court is potentially on notice that potentially the lay litigant has not looked into potential cases.
“I do think it would be useful having that proactivity and is also certainly consistent with our duties under the client care rules, but it is not a requirement of the court at present.
Learn more about Generative AI in The Law Association’s upcoming webinar ‘Navigating the Law with Generative AI (2026)’.
0 Comments