People will leave the Victoria Supreme Court in Melbourne on August 15th, 2025. Photo Credit: AP
A senior Australian lawyer apologised to the judge for submitting submissions in a murder case that included false citations and judgments of nonexistent cases created by artificial intelligence.
The Victorian Supreme Court's blunder is another of a series of AI-induced accidents in justice systems around the world.
Defense attorney Rishi Naswani, who holds the honorable legal title of King's attorney, has received “full responsibility” for submitting false information in the case of a teenager charged with murder, according to court documents. Associated Press Friday (August 15, 2025).
“We are deeply sorry and embarrassed about what happened,” Naswani told Judge James Elliot on Wednesday (August 13, 2025) on behalf of the defense team.
The error generated by the AI was 24 hours late to resolve the case Judge Elliott hoped to close on Wednesday (August 13, 2025). Judge Elliott determined on Thursday (August 14, 2025) that Naswani's client, who cannot be identified because he is a minor, was not a murder charge due to mental illness.
“The way these events unfold is insufficient due to the risk of understatement,” Judge Elliott told his lawyer on Thursday (August 14, 2025).
“The ability of a court to rely on the accuracy of a lawyer's submission is fundamental to justice control,” Judge Elliott added.
The fake submissions included a citation produced from the speech to the state legislature and a quote from the allegedly nonexistent case from the Supreme Court.
The error was discovered by Judge Elliott's companions. Judge Elliott's companions were unable to find the case and requested that the defense attorney provide a copy.
The lawyer admitted that the citation was “nonexistent” and the submission contained “fictional citations,” court documents state.
The lawyer explained that he confirmed that the initial citation was accurate and that he incorrectly assumed the other citations were correct as well.
The submission was also sent to Prosecutor Daniel Pordedu. Daniel Poldedou did not check the accuracy. The judge noted that the Supreme Court last year issued guidelines on how lawyers can use AI.
“Using artificial intelligence is not acceptable unless the product of its use is independently and thoroughly verified,” Judge Elliott said.
Court documents do not identify the generative artificial intelligence systems used by lawyers.
In a comparable US case in 2023, a federal judge fined two lawyers and the law firm $5,000 after ChatGPT was blamed for fictitious legal investigations in an aviation injury claim.
Judge P Kevin Castell said they acted maliciously. However, he believes they or others have taken an apology and corrective action in explaining why stricter sanctions are not needed to ensure that artificial intelligence tools do not once again encourage them to create fake legal history in their arguments.
Later that year, the more fictitious court decision invented by AI was cited in legal documents filed by lawyers for Michael Cohen, former private lawyer for US President Donald Trump.
Cohen took responsibility for saying that the Google tools he used for legal investigations didn't realize that so-called AI hallucinations were possible.
Published – August 15, 2025 04:48 PM IST
