
G’day mates.
The latest installment of “Lawyers Royally Bungling with AI” features an Australian attorney who apparently decided that verifying case citations was just too 2024. After prompting the consumer-facing AI to do some research (bad idea), and putting that into a filing (worse idea), he decided to throw another shrimp on the barbie and submit the documents filled with references to phony cases to the court (worst idea).
A reminder that every time you think this story has been hyped enough that lawyers won’t do it again… they go ahead and do it again.
From The Guardian:
An Australian lawyer has been referred to a state legal complaints commission, after it was discovered he had used ChatGPT to write court filings in an immigration case and the artificial intelligence platform generated case citations that did not exist.
In a ruling by the federal circuit and family court on Friday, Justice Rania Skaros referred the lawyer, who had his name redacted from the ruling, to the Office of the NSW Legal Services Commissioner (OLSC) for consideration.
Presumably the judge said, “That’s not a case… this is a case.”
This legal eagle, whose name has been thoughtfully redacted to protect the technologically inept, was handling an immigration appeal when he submitted filings in October 2024. These documents were so compelling that they cited and even quoted from tribunal decisions that simply didn’t exist. When confronted, the lawyer confessed to using ChatGPT for research, admitting he didn’t bother to verify the AI-generated information.
Despite the temptation to turn this into a story about technology run amok, this is still fundamentally a matter of human laziness. Just as a lawyer shouldn’t mindlessly blockquote the memo the summer associate slapped together on their way to happy hour, any lawyer using generative AI retains the obligation to check the final product for accuracy. And, in the case of ChatGPT legal research, it’s more like the memo the summer associate slapped together on their way back from happy hour.
He attributed this lapse in judgment to time constraints and health issues. Maybe that’s true. But the most troubling detail from this story is that this happened in an immigration case. One of the more noble selling points for generative AI is the hope that it could expand access to justice by streamlining practices like immigration. Filevine, for instance, offers some really slick AI-assisted immigration tools — all of which are very different than turning over briefing to a free chatbot. Yet with this promise comes the risk that practice areas that serve the most vulnerable will be the most likely to get shortchanged out of attorney judgment. That paying client will get proper attention while the pro bono matter gets churned out by lightly checked AI.
Again, that might not have been the case here, but this is the new frontier for AI screw-ups. More lucrative practices are going to get lawyer attention and the benefit of high quality AI tools crafted specifically for the legal profession with all the safeguards that requires. And it’s going to be the lower income practices that give rise to future embarrassing cases.
And — in matters like immigration — potentially tragic ones.
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news. Joe also serves as a Managing Director at RPN Executive Search.