← Back to feed

Australia’s Fair Work Commission has taken the rare step of ordering a self-represented litigant to pay partial legal costs after finding he used ChatGPT as a “quasi-legal advisor” — and then kept filing despite the chatbot’s own assessments pointing to a hopeless case.

Sadnan Khan, dismissed from ALDI and seeking reinstatement, used a paid ChatGPT subscription to navigate the commission’s process. The commission’s deputy president Michael Easton did not object to AI use in principle. He objected to the outcome: Khan had ignored repeated warnings from both the tribunal and, according to the decision, from his own AI tool.

“If Mr Khan had properly read his own AI generated replies … he would have known that his case was doomed,” Easton wrote in the ruling.

Khan was ordered to pay $1,230 toward ALDI’s legal costs — described by the commission as a rare sanction, reserved for conduct that forces the opposing party to incur unnecessary expense.

What Went Wrong

Khan told the ABC he used ChatGPT to interpret the cases ALDI cited in its defence, describing it as a navigation aid for someone unfamiliar with Australian legal process. He acknowledged the AI gave him advice “not the Aussie [court] way” and admitted to submitting documents that still contained ChatGPT chat instructions — left in accidentally.

The commission did not treat that as dishonesty. It treated it as unreasonable conduct: Khan continued to prosecute a claim after the evidence — including, apparently, his own AI-generated analysis — indicated it had no chance.

“The main problem, which I realised after the hearing, was that I was thinking oranges were apples,” Khan said. His next step: use “two or three” AI agents simultaneously — Claude and ChatGPT among them — to get competing views.

The Other Side of the Ledger

The same ruling that condemned Khan’s AI use acknowledged an Australian-first from the opposite end of the spectrum. Gregory Baker, a computer science lecturer at Macquarie University, became the first person to successfully challenge casual employment laws using a coordinated team of AI agents. Baker has publicly advocated mandatory AI use to his students and described it as essential preparation for professional work.

The commission’s language on AI in general was measured. It denied that AI use by applicants was “inherently problematic” and said users who approached it critically as “one tool among many” had produced genuinely meritorious claims that might otherwise never have been filed.

Regulatory Context

Australia’s Fair Work Commission and Administrative Review Tribunal have both tightened GenAI disclosure rules in 2026. The broader pattern is not unique to Australia: UK Employment Tribunal filings surged roughly 100-fold as AI-assisted litigants flooded the docket, with cases expected to queue until 2030. Legal regulators in multiple jurisdictions are converging on the view that AI-generated submissions require disclosure, verification, and professional oversight — the same due diligence that applies to any other tool used to prepare court documents.

The Khan case adds a data point that is harder to quantify: how much of the AI legal advice surge represents genuine access-to-justice improvement versus an increase in unmeritorious claims that waste tribunal time and opponent resources. The $1,230 costs order is small. The signal it sends — that tribunals will hold AI-assisted litigants to the same standard of reasonable conduct as any other party — is not.

Key Numbers

• Costs ordered: AUD $1,230 (toward ALDI’s legal costs)
• Commission: Fair Work Commission, Deputy President Michael Easton
• AI tool used: Paid ChatGPT subscription
• Outcome of unfair dismissal claim: Dismissed with costs
• Comparative: Gregory Baker (Macquarie) — first successful casual employment challenge using AI agents, same tribunal period