A Sydney man charged over a data breach of a New South Wales courts website allegedly used OpenAI's ChatGPT to code software for the mass downloading of documents, a court has reportedly heard.

Christopher Duff, 40, appeared in Sydney's Downing Centre Local Court this week for a three-day hearing, after pleading not guilty to four counts of accessing restricted data over almost 9,000 sensitive documents police said were accessed in early 2025.

Officials previously said the documents included apprehended violence orders, details of minors, and other forms held by the NSW Department of Communities and Justice (DCJ) portal known as the NSW Online Registry.

Duff was charged in April 2025 when police executed a search warrant at a home in Sydney's east and seized two laptops, before later being granted bail.

Prosecutors this week alleged he used generative AI chatbot ChatGPT to create scripts of scraper-style software in the Python coding language to allow him to bulk download data, ABC News reported.

Police prosecutor Sergeant Andrew Maldigri reportedly told Judge Lisa Graham, "It means a user has used computer coding to obtain data that they weren't authorised to.

"That's, in a nutshell, what this matter is about."

Prosecutors are reportedly seeking to rely on Duff's conversations with ChatGPT as evidence, in a case which is believed to be among the first of its kind in Australia.

Duff's defence counsel Daniel McMahon reportedly told the court there were an estimated 10,000 or more pages of proposed exhibits.

Allegedly sought 'coaching' from ChatGPT

Prosecutors have reportedly alleged Duff used ChatGPT to seek legal advice before he was charged, and when his login for the NSW Online Registry was blocked.

Maldigri reportedly told the court Duff sought out "coaching" from ChatGPT about what to say to police, before he declined to answer their questions in a formal interview.


Police allege Christopher Duff used ChatGPT to generate software code for the mass downloading of documents. Image: Shutterstock

McMahon reportedly said his client had asked the AI chatbot whether or not "what was done was unauthorised", or a crime.

On Wednesday he told the judge there may not be disputes over whether his client was responsible for some of the alleged conduct, ABC News reported.

McMahon reportedly suggested prosecutors needed to prove that Duff knew he was accessing documents while unauthorised, and not just being reckless.

The court reportedly heard Duff was attempting to help a friend who was undergoing bankruptcy proceedings.

The case has been adjourned for two weeks, ahead of what could be a longer hearing.

GenAI in legal proceedings

Police prosecutors reportedly told the judge that using generative AI as evidence, and to prove motive, was yet to be raised in Australian courts.

They reportedly mentioned examples from the United States, including of a man in New York who unsuccessfully tried to prevent records of his usage of Anthropic's Claude AI chatbot from being used against him.

In January 2025, US authorities said the driver of a rented Tesla Cybertruck who died when the vehicle exploded in Las Vegas had used ChatGPT to help plan the blast.

Australians have previously gotten into trouble for using AI chatbots in submissions to the Fair Work Commission, including a worker whose case was thrown out over inaccurate information, and another who was ordered to pay costs after repeatedly relying on incorrect AI legal advice.

But a computing academic successfully used ChatGPT to win a historic case in August after asking for advice to defeat his employer's human lawyers.