An Australian worker has been ordered to pay $1,230 in costs after repeatedly relying on faulty AI-generated legal advice to pursue an unfair dismissal claim he was never eligible to make.

The use of artificial intelligence platforms such as ChatGPT has been a bugbear for the Fair Work Commission (FWC) this year, creating a sharp uptick in caseload and an increase in filings with clear mistakes and little hope of succeeding.

On the flip side, the advent of generative AI has helped Australian workers know their rights and take issues to the FWC, as evidenced last week when a Macquarie IT lecturer won a case against his university’s high-power lawyers using AI.

But the use of AI at the FWC did not go as well for a former Aldi employee, who in March took the company to the FWC alleging unfair dismissal.

Under FWC rules, a worker must have been employed by a company for at least six months to be eligible to make an unfair dismissal application

The relevant date for calculating the six-month minimum is when the employee is notified of their dismissal.

When the Aldi worker was told that they would be let go, they were three days short of the six-month minimum period, meaning they were not allowed to make a claim with FWC.

An AI-motivated ‘hopeless’ case

Fair Work Commission deputy president Michael Easton said the dismissal date was never in dispute by either party, but that the worker’s filings, which he said were clearly AI generated, continued to rely on the wrong date.

“Ever since the minimum period problem was first raised, [his] AI-generated materials have focused on a different and relevant date, being the date the dismissal took effect,” Easton said in his findings.

The materials filed with the FWC contained AI drafting notes including: “Here is your **final clean copy-paste version” and “Got it. I have updated your **Outcome Requested** section.”

The deputy president quickly came to the view that the Aldi worker’s case was “hopeless” and wrote to him informing him of this, encouraging him to discontinue the case.

In the email, the worker was also warned that he may be on the hook for Aldi’s legal costs if he continued with the case despite these warnings.

The worker replied with another “AI-generated” email which continued to rely on the wrong date and repeated “irrelevant AI-generated arguments”, FWC heard.

Easton said that while the filings weren’t “complete AI slop”, they contained a wealth of “irrelevant” legal material.

AI as a ‘quasi-legal advisor’

Despite further emailed warnings that he had a “very strong likelihood” of not winning the case and Aldi had a “very strong argument” of getting their costs paid, the worker sent another two AI-generated emails using the false information.

“It is quite possible that [he] adopted AI as his quasi-legal advisor and then blindly submitted AI-generated outputs as his own material,” FWC said.

“However, it was [his] responsibility to properly assess for himself whether he was eligible to make his unfair dismissal claim and it was his responsibility to ensure that whatever he submitted to the Commission addressed the concerns raised.”

It wasn't until the actual hearing that the worker admitted he hadn’t worked at Aldi long enough to make the claim, and he discontinued the application.

Under FWC rules, costs orders are very rare, as the commission is designed to be accessible and easy for workers to represent themselves, without fears of having to pay high legal costs.

But if the commission is satisfied legal costs were incurred due to an unreasonable act or omission, an order can be made for the applicant to pay them.

Easton ultimately found the worker should pay some of the costs that Aldi “wasted in preparing for and attending a hearing that should never have taken place”.

This amount was calculated as the maximum amount allowed that was incurred after the worker was warned that there was a chance they’d have to pay costs.

“There has been no winner in this matter,” Easton said in his findings.

“[He] was never eligible to make his unfair dismissal claim and is now $1,230 worse off than he was before he started.

“The losses for both parties were utterly preventable."