The Supreme Court has cautioned that relying on false AI citations “may in serious cases amount to obstruction of justice or contempt,” as recent cases in the High Court and Tenancy Tribunal reinforce judicial concerns about generative AI use in legal contexts. Duncan Cotterill lawyer Jeremy Ansell told The Press that AI use in legal contexts was only increasing, and courts would be unable to stop people from using it, calling it “inevitable.” He noted AI “lowers the bar to access [to justice],” but added that clients are receiving “enormous submissions that are full of AI jargon.” In a Tenancy Tribunal case, Christchurch landlord Inspire Property Management claimed $95 for cleaning and a missing pillow from tenant Anna Kakinuma, who applied for $11,500 of compensation. Adjudicator J Greene said Kakinuma’s 17-page submissions, referring to a “severe pattern of professional negligence, harassment, and attempted fraud,” appeared AI-prepared, rejecting both claims in late June. The same month, Associate Judge Dale Lester did not allow Christchurch director Christopher Roberts to represent his company in a $1.1 million earthquake repair claim after he used AI in an application that seemingly made up an authority. Ansell said unclear submissions could cause disputes to run longer, raising costs. The main theme from courts, including AI guidelines from 2023, was that AI users are responsible for what they submit.