New Zealand Judge Reduces Sentence After AI-Written Apology Raises Doubts

Compiled by the editorial desk with reference to official court transcripts, public reports from The New Zealand Herald, and coverage by The New York Times.

A New Zealand judge has handed down a 27-month prison sentence to a woman convicted of arson, but the case has drawn attention for a different reason: the apology letters she submitted were generated by artificial intelligence, leading the judge to question the sincerity of her remorse.

District Court Judge Tom Gilbert in Christchurch noted during sentencing that he had used two AI tools to test the letters, which were addressed to him and to the victims. According to a transcript reviewed by The New York Times, Gilbert said, “Out of curiosity I punched into two AI tools ‘draft me a letter for a judge expressing remorse for my offending.’” He added, “It became immediately apparent that these were two AI-generated letters, albeit with tweaks around the edges.”

The defendant, who pleaded guilty to arson and other charges, also bit a police officer while in custody and reportedly told the officer she had AIDS, taking “delight” in the disclosure, as first reported by The New Zealand Herald.

Remorse is a recognized factor in sentencing, often allowing for a reduced term. However, Gilbert made clear that a computer-generated letter does little to demonstrate genuine regret. “But certainly when one is considering the genuineness of an individual’s remorse, simply producing a computer-generated letter does not really take me anywhere as far as I am concerned,” he said.

Her lawyer had sought a 10% reduction in the sentence, but Gilbert granted only 5%, calling even that “reasonably generous.” The defendant will serve 27 months in prison.

AI’s Troubled Track Record in Courtrooms

This incident is not isolated. Courts have seen multiple cases where AI tools produced hallucinated content in legal filings, including fabricated case law and incorrect citations, leading to judicial reprimands and internal crises at law firms. In October, a defense lawyer who had used AI in court filings submitted an apology that was also AI-written, initially denying it before later admitting the use.

Judges are increasingly alert to the presence of AI-generated material, and this case underscores the growing tension between technological convenience and legal authenticity. While AI can assist in drafting, its use in matters of personal accountability, such as apologies, may undermine the very purpose of the act.

As courts navigate this new terrain, the expectation is that legal professionals and defendants alike will need to exercise greater caution, ensuring that technology does not replace the human element essential to justice.