Fair Work Commission condemns 'plain wrong' AI legal advice

Aug 30, 2026 05:20 PM - 2 weeks ago 7

A sacked ALDI worker condemned by the Fair Work Commission for utilizing "plain wrong" artificial intelligence (AI) proposal successful his grounded bid to reverse his dismissal says he now plans to usage "two aliases three" different AI agents to thief him entreaty against the decision.

Sadnan Khan was ordered to salary $1,230 towards the supermarket giant's ineligible fees aft a landmark Fair Work Commission determination recovered he utilized AI arsenic a "quasi-legal advisor" to situation his dismissal, contempt repetition warnings his lawsuit had "no important prospects of success".

"I person taken the very uncommon measurement of ordering him to salary immoderate of the employer's ineligible costs, because his unreasonable behaviour caused his erstwhile employer to incur those costs unnecessarily," Fair Work Commission lawman president Michael Easton said successful the tribunal's determination past week.

"If Mr Khan had decently publication his ain AI generated replies … he would person known that his lawsuit was doomed."

An ALDI supermarket sign

A sacked ALDI worker challenged his dismissal successful the Fair Work Commission. (ABC News: Sarah Maunder)

Mr Khan told the ABC he utilized a paid type of ChatGPT to thief him decipher ineligible cases the tribunal and ALDI had raised.

"They're quoting maine cases, correct … and I arsenic not a ineligible characteristic person to spell location to do the study," he said.

"The main point AI suffers is they do things not the Aussie [court] way."

Mr Khan — primitively from Bangladesh — said he was relying connected AI proposal to understand really to navigate the system, but admits he made mistakes — including forgetting to region AI-chat instructions from his submissions.

"The measurement we pass things and the ineligible measurement of connection … things are different," he said.

Generative AI sees cases surge

The Fair Work Commission is designed to assistance self-represented litigants, but a study commissioned by the workplace umpire showed a drastic uptick successful cases.

Mr Khan contends comprehension barriers affected the prompts he fed his AI agent, but said he had now learned from the experience.

"The main problem, which I realised aft the hearing, was that I was reasoning oranges were apples," he said.

"Now I'm going to usage adjacent clip a substance of Claude and ChatGPT … and 2 aliases 3 [others] to [get a] substance of different views."

The tribunal acknowledged AI was improving entree to justness for applicants pinch genuine, meritorious claims "who mightiness different person been deterred".

It denied AI usage by applicants was inherently problematic and said blase users had approached it critically arsenic "one instrumentality among many".

ChatGPT 'terrible arsenic oracle' 

Among that class is Gregory Baker, a machine subject teacher from Macquarie University who this period became the first personification to successfully situation casual employment laws utilizing a squad of AI agents.

"I utilized to beryllium celebrated for saying to my students, 'if you don't usage AI successful my class, you will fail',"

Mr Baker said.

"When you're school a people connected AI, it makes sense."

A teacher successful beforehand of seating and students

Macquarie University teacher Gregory Baker is the first personification to successfully situation casual employment laws utilizing a squad of AI agents. (ABC News: Abubakr Sajid)

Mr Baker's Fair Work declare was calved erstwhile he asked ChatGPT for profession readying options aft the assemblage denied his petition to beryllium converted from a sessional to imperishable employee.

"I treated it fundamentally arsenic a package improvement project," he said.

"I created a repository for my root codification and my programs, which were my [tribunal] filings and came up pinch a build process that it checked.

"Are each the citations correct? Is everything logically coherent? I made judge location was a process that it went through.

"That evidently has tremendous implications for really institutions operate, for really the ineligible systems work."

But he said asking ChatGPT arsenic a "kind of oracle" pinch nary discourse aliases accusation resulted successful "a unspeakable job."

A machine subject teacher astatine a desk

Computer subject teacher Gregory Baker encouraged his students to usage AI. (ABC News: Abubakr Sajid )

Mr Baker estimated that location was astir a five-year model betwixt the AI frontier, which had "good framing and prompting" compared pinch the slightest sophisticated, and often free, models, which could thrust inequality.

"There are a batch of group for whom hundreds of dollars a period connected AI is wholly retired of the question," he said.

Mr Baker did not realise he was among the first cases testing caller worker prime laws introduced past twelvemonth until a week aliases 2 earlier the hearings, wherever he was up against a lawyer and barrister representing the university.

He said he utilized ChatGPT to "grill" him to hole for cross-examination astatine the tribunal hearing, but admitted, "things did not spell the measurement I was expecting."

"Of course, it's a immense disadvantage successful immoderate oral hearing,"

he said.

How tin courts, tribunals negociate AI litigants?

AI tin assistance litigants champion erstwhile the personification understands the discrete ineligible rumor they are querying, Genevieve Grant, a rule professor from Monash University, said.

"If those prompts are inexpert, past the results will beryllium arsenic well," she said.

An world pinch a precocious bun and reddish lips

Monash University rule professor Genevieve Grant says inexpert AI prompts get inexpert results. (Supplied)

"Most organization members are alert of this benignant of spot of AI now wherever it has a benignant of sycophantic inclination … and whitethorn successful truth promote group to judge that location are legs successful a declare they're making erstwhile that whitethorn not successful truth beryllium the case."

Professor Grant said the tribunal's reasons successful Mr Khan's lawsuit showed Mr Khan seemed to reside the incorrect ineligible mobility contempt attempts to put him connected the correct course.

"Courts and tribunals person a constricted scope of devices that are disposable to them to nonstop messages to the organization … 1 of those devices is simply a costs penalty," she said.

"I deliberation what we're going to spot is courts and tribunals utilizing an expanding array of devices to negociate this, and we mightiness spot a ramping up of standards astir the sorts of expectations that courts and tribunals have, really overmuch latitude they supply to litigants.

"Really we're talking astir the sorts of measures that mightiness deter immoderate group from bringing a declare aliases weeding retired those claims astatine an earlier shape than we're often utilized to."

New template for AI-dependent litigants 

The Fair Work Commission precocious established a template for AI-dependent litigants, encouraging them to transcript and paste prompts into their AI agent.

Guidance changes announced by the tribunal connected Monday will besides require applicants from October 20 this twelvemonth to disclose immoderate AI use, verify facts, authorities and hyperlinks used, and corroborate witnesser evidence. 

It has flagged consequences for those who do not comply.

Mr Khan, whose lawsuit centred connected whether he was eligible to situation his dismissal, said he was not deterred by the anticipation of a much costly punishment if he followed done pinch his scheme to entreaty the determination successful his case.

His model to entreaty will extremity earlier the Fair Work Commission changes return effect.

"Let's spot really it goes, I will look into my options,"

he said.

"AI besides helped maine successful the past for a fewer different matters."

Macquarie University has until Wednesday adjacent week to entreaty against the determination successful Mr Baker's case.

More