News
Australian Tribunal Orders Fired ALDI Worker to Pay Costs Over ChatGPT Reliance
Australia's Fair Work Commission ordered a former ALDI employee to pay AU$1,230 in legal costs after he treated ChatGPT as a legal adviser in his unfair dismissal appeal. The tribunal also revealed that signs of AI use now appear in 50 to 75 percent of the cases it reviews.
Contents
On August 29, 2026, Australia's Fair Work Commission ordered a former ALDI employee to pay AU$1,230 in legal costs incurred by his employer. The reason was the man's decision to rely on the paid version of ChatGPT as what the tribunal called a "quasi legal adviser" in his appeal against dismissal.
The Case of Sadnan Khan
Sadnan Khan, an ALDI employee originally from Bangladesh, challenged his dismissal before the Fair Work Commission, Australia's employment tribunal. Instead of consulting a lawyer, he used the paid version of ChatGPT to understand the legal precedents cited by the tribunal and by the retailer itself.
The problem was that the documents filed with the commission contained fragments generated by the chatbot that Khan had not removed, including the instructions addressed to the AI itself. Commission Vice President Michael Easton left no doubt about the quality of that legal assistance.
If Mr Khan had properly read his AI-generated responses, he would have known his case was doomed to fail - Michael Easton, Vice President of the Fair Work Commission
The Scale of the Problem
Khan's case is not an isolated one. According to data cited by the tribunal and Australian trade media, nearly 40 percent of people filing applications with the Fair Work Commission already use generative AI to prepare their documents, and 77 percent of them turn to ChatGPT. In mid-2026, commission members were recognizing signs of AI use in 50 to 75 percent of the cases they reviewed.
That prompted the tribunal to introduce new, formal rules for handling documents prepared with the help of artificial intelligence. Until now the commission had responded to the problem case by case; now it will have uniform guidelines applying to all applicants.
New Rules Starting in October
Starting October 20, 2026, anyone who uses generative AI tools such as ChatGPT, Claude, Copilot or Gemini to prepare submissions to the Fair Work Commission will have to meet three conditions. First, disclose the use of AI and explain how the tool was used. Second, independently verify the accuracy of all facts, evidence, statutes and case citations. Third, in the case of witness statements, confirm that the content actually reflects what the witness knows.
Lawyers and business representatives appearing before the commission will face an additional obligation: they must include hyperlinks to the rulings they cite, making it easier to quickly verify whether the precedents referenced actually exist and say what the submission claims.
Consequences of Non-Compliance
Documents that fail to meet the new requirements may be given less evidentiary weight by the commission or rejected outright. The tribunal has also reserved the right to order a party to pay the other side's costs, as happened in Khan's case, and in extreme situations to dismiss a case entirely. Knowingly submitting false AI-generated information presented as one's own findings can carry a penalty of up to 12 months in prison.
The rules apply to all individual cases handled by the commission, including applications alleging unfair dismissal and cases involving general protections for workers' rights, the categories where people without their own lawyer most often turn for help.
The Risk for Unrepresented Litigants
Khan's case illustrates a pattern that keeps repeating in similar cases around the world: people without access to paid legal help treat chatbots as a cheap substitute for a lawyer, not realizing that a conversation with an AI is not covered by attorney-client privilege and its content can be disclosed to the other party in the proceedings. On top of that comes the risk of hallucination, meaning the generation of nonexistent cases and statutes that then end up in official court filings.
For readers in Poland, the case carries significance beyond Australia. Similar questions about the limits of using AI in judicial and administrative proceedings are coming up increasingly often in Europe too, and case law from English-speaking countries, where the number of AI-related cases is growing fastest, usually sets the direction for regulation elsewhere later on.
The Fair Work Commission has said it will monitor how effective the new rules prove to be after they take effect in October and will tighten them if needed. For now, Khan's case remains the starkest example yet of how costly it can be to trust a chatbot when the outcome of a legal case is at stake.
