Australian lawyer reported to regulator for filing fake Gen AI cases in court

Legal Insider ^ | October 15, 2024 | Caroline Hill

[xxqqzz]

A lawyer in Melbourne, Australia has been referred by a federal circuit judge to the Office of the Victorian Legal Services Board and Commissioner after filing a summary of authorities that do not exist, which he prepared using Gen AI assistant Leap LawY.

The case, which happened in the summer but we first became aware of thanks to coverage on 10 October by The Guardian newspaper, was heard by Justice Amanda Humphreys. Subsequent research confirms that another lawyer has fallen foul of Gen AI hallucination, and filing case law without verifying the accuracy of the information.

Justice Humphreys adjourned divorce proceedings in July after the fake citations came to light, ordering the anonymous solicitor ‘Mr B’ of ‘D Law Firm’ to email the court with reasons why he ought not be referred to the regulator for his conduct.

In an order dated 27 August, Justice Humphreys ruled that Mr B (who she ordered remain anonymous) must be investigated, after he admitted to not reviewing but still filing the fictitious output generated by Leap LawY.

In his defence, Mr B said that he didn’t fully understand how the research tool worked and offered to pay the costs of the adjourned July hearing. He also submitted personal details around the stress that the July hearing had caused him.

Justice Humphreys referred back to the US case of Mata v Avianca, in which Steven Schwartz famously filed case law that didn’t exist after using ChatGPT. She also referred to the guidelines from the Supreme Court of Victoria and County Court of Victoria, which say that parties who are using AI tools in the course of litigation need to understand how they work; the use of AI programs mustn’t mislead other parties and their use should be disclosed;

[xxqqzz]

Another lawyer got in trouble using AI tools, which generate fake cases. The first case of this was in federal court in NYC where lawyers used Chatgpt to come up with legal arguments. Chatgpt made up fictitious cases, which the lawyers didn't check. The arguments also didn't make sense. That NYC case involved a Hispanic guy suing a Columbian airline saying he sustained severe injuries when a flight attendant hit his knee with a tray. The case was apparently passed the statute of limitations, and the airline was bankrupt, in addition to the injury case being questionable. I am not a lawyer, but know enough about this sort of thing to comment. The lawyers in NYC were fined, but their reputations damaged. A lawyer in Colorado was suspended from practicing for this. Complete stupidity, dishonesty, and misuse of AI tools.