Fair Work warns litigants against using ChatGPT

The workplace tribunal has dismissed a worker’s claim that relied on ChatGPT for legal advice as being ‘hopeless and unnecessarily wasting the resources of the commission’.

Fair Work found a worker’s claim that relied on ChatGPT for advice to be ‘hopeless’. Picture: Kiichiro Sato

The Fair Work Commission has warned of the dangers of litigants relying on artificial intelligence for legal advice, dismissing a worker’s claim that relied on ChatGPT for advice as “hopeless” and a waste of the workplace tribunal’s resources.

Sydney man Branden Deysel sought an extension of time to file an unfair dismissal claim against his former employer, Electra Lift Co. Claims are supposed to be filed within 21 days of a dismissal but Deysel resigned from the company in October 2022, and filed his claim 919 days later.

Mr Deysel confirmed to the commission that he had used ChatGPT, which advised him his former employer had contravened workplace laws and he should file the claim.

Lawyers said using ChatGPT would be tempting for commission applicants, many of whom were self-represented litigants.

Michael Byrnes, partner at law firm Swaab, said the commission deci­sion illus­trated the risks and dan­gers of using Chat­G­PT, or oth­er AI appli­ca­tions or mod­els, for the prepa­ra­tion of commission appli­ca­tions and responses.

“While it made lit­tle dif­fer­ence in this case, it is only a mat­ter of time before an oth­er­wise mer­i­to­ri­ous appli­ca­tion or response is under­mined by the mis­use of AI apps or mod­els,” Mr Byrnes said.

In his decision this month, commission deputy president Tony Slevin said it was clear Mr Deysel had used ChatGPT, given the deficiencies in his ­application, which failed to address the ­matters required to make good his claims that the Fair Work Act had been contravened.

Mr Deysel’s application included an extract from advice given by Chat GPT which was that various employment and other statutory obligations had been contravened by his former employer, and he should pursue legal action.

“I can see no basis for this advice,” Mr Slevin said, before dismissing the claim. “Chat GPT also advised Mr Deysel to consult a legal professional or union representative to determine the appropriate course of action. He did not do so.

“(He) simply followed the suggestion made by Chat GPT and commenced the ­proceedings. The circumstances highlight the obvious danger of relying on artificial intelligence for legal ­advice.

“The result has been Mr Deysel commencing proceedings that are best described as hopeless and unnecessarily wasting the resources of the commission and the respondent in doing so.”

Mr Byrnes said the risks and dan­gers of using ChatGPT to file applications was par­tic­u­lar­ly acute in a juris­dic­tion like the commission, where prin­ci­ples of fair­ness, which require val­ue judg­ments, needed to be care­ful­ly con­sid­ered.

“The work of the FWC does not lend itself to a for­mu­la­ic approach, slav­ish to pre­vi­ous ­­­cas­es that may osten­si­bly have sim­i­lar facts. It is a trite propo­si­tion that each case turns on its own circumstances,” he said.

Mr Byrnes said courts and tri­bunals fre­quent­ly deal­ing with self-rep­re­sent­ed lit­i­gants needed to be acute­ly aware of the pos­si­bil­i­ty ChatGPT was being used. “If it is, and it leads to a sound case being poor­ly argued or pre­sent­ed, that is the fault of the par­ty seek­ing to rely on the tech­nol­o­gy.

“The FWC should give no con­ces­sion or lat­i­tude to any par­ty for hav­ing made that decision.

“AI may have a role to play in FWC pro­ceed­ings but, at this stage in its devel­op­ment, it needs to be used judi­cious­ly and its out­put treat­ed with a healthy degree of scepticism.”