A Western Australian worker has learnt the hard way that you can’t rely on AI to prepare your Fair Work unfair dismissal claim. Her AI-generated submissions cited case law that “simply did not exist” as well as two irrelevant cases. She had told the Fair Work Commission that she had “checked” all of her submissions, but this declaration was found to be “misleading.” The Commission said her responses under questioning showed “a pattern of evasiveness,” leading it to conclude “she had not actually read” her submissions. Her Fair Work claim was ultimately dismissed, having “wasted significant time” for the Commission and the opposing party.
In this article, we look at the events of this Fair Work unfair dismissal case: Sarah Byrne v CXC Corporate Services Pty Ltd [2026]. Later, we’ll look at a recent case where Fair Work banned a worker from using AI to make any further complaints about her boss.
WA-based Sarah Byrne was engaged by CXC Corporate Services to work as a site senior health and safety advisor for its client, WSP Australia. She was employed as a casual under a labour hire arrangement. On 14 April 2026, WSP terminated the assignment Ms Byrne was undertaking, and she was removed from the site. She subsequently lodged a Fair Work unfair dismissal claim, representing herself during proceedings.
CXC initially argued to the Fair Work Commission Ms Byrne had not been dismissed, given she had been offered other assignments which she had declined. But the company later abandoned this objection. Instead, CXC lodged a jurisdictional objection to Ms Byrne’s Fair Work unfair dismissal claim. The firm argued that her annual rate of earnings was above the Fair Work unfair dismissal high-income threshold, in circumstances where she was not covered by a modern award or enterprise agreement. Ms Byrne maintained she was covered by the Miscellaneous Award 2020. She also argued that her casual earnings could not be calculated into an annual rate for the purpose of determining whether she exceeded the high-income threshold.
Early in proceedings, the Fair Work Commission “sounded a note of caution” to Ms Byrne about using AI to inform her case. Despite this, Ms Byrne relied on generative AI to prepare her submissions. She attached a declaration at the end of each section stating that she had used AI to help prepare her submissions.
The declaration said that she had “checked” the document to ensure all facts, cited case law and legal arguments were “accurate, relevant and appropriate to this case.” However, the Commission found that this declaration “proved to be misleading.”
In arguing that her role was covered by the Award, Ms Byrne cited three cases. The first, Ponce v DJT Staff Management Pty Ltd, was a real case. However, the Fair Work Commission found it had “no relevance” to the award coverage issue. When this was put to Ms Byrne, she responded in a way the Commission described as beginning “a pattern of evasiveness.” This led to the conclusion that “she had not actually read the case.”
The second case cited, Rahman v Active Labour Pty Ltd, could not be found despite the Commission “scour[ing] several legal databases.” CXC’s legal counsel also confirmed it could not find the case either. The third case was referred to by Ms Byrne only as the “CXC / Calibre line.” Upon further questioning, this case was actually Mr Allan Gabriel v Titan Recruitment Pty Ltd T/A Titan Recruitment. The case concerned an extension of time for a Fair Work general protections claim. It therefore had “no relevance whatsoever to the matter at hand,” the Commission said.
Given one case was fake and two were irrelevant, the Commission noted that it appeared Ms Byrne had not actually read the cases. It said that she “wasted significant time” for both the Commission and CXC’s legal representatives. The Commission also noted that similar “presumably AI-generated” errors also affected case citations in her later submissions regarding her earnings.
The jurisdictional objection to Ms Byrne’s Fair Work unfair dismissal claim was heard by the Fair Work Commission in July 2026. It threw out her claim based on the following findings:
In Mrs Deidre De Bruyn [2026], a Queensland caretaker who lodged a Fair Work stop bullying claim was banned by the Fair Work Commission from using AI tools in future communications with her employer. This came after the Commission found her AI-generated correspondence was so “lengthy, wide-ranging” and “accusatory” that it likely worsened the very dispute she was trying to resolve.
Deidre de Bruyn was co-director of Laro Real Estate, along with her husband Mattheus de Bruyn. The company held the caretaking contract for the body corporate for the Sanctuary residential complex in Queensland’s Sunshine Coast region. The contract provided for the de Bruyns to reside on-site and maintain the property’s gardens and amenities.
A dispute over the scope of the caretaker’s duties arose soon after the contract began in 2023. This significantly deteriorated the relationship between the de Bruyns and the body corporate. In September 2024, Ms de Bruyn applied for a stop-bullying order with the Fair Work Commission. That application was resolved by a settlement agreement reached at conciliation in November 2024.
Approximately 12 months later, Ms de Bruyn made a second application. She alleged that she had continued to be bullied by the Sanctuary body corporate committee. She submitted over almost 400 pages of material to support her claim of bullying. Ms de Bruyn alleged that the committee repeatedly refused to clarify her duties, excluded her from processes affecting her role and published misleading information about her performance. She also claimed that she had been subject to intrusive monitoring of her movements and appearance.
Read more: Miner’s ‘Egregious’ Lewd Banter Justified Dismissal
The Fair Work Commission found several instances of unreasonable conduct directed at Ms de Bruyn. In February 2026, the chairman of the committee spent “a couple of hours googling” a proposed representative Ms de Bruyn wished to appoint. He later shared derogatory comments about both the representative and Ms de Bruyn with the committee. The following week, the chairman kept a diary of observations regarding Ms de Bruyn’s appearance and activities on the premises, in connection with whether it appeared she was working. He later shared this to the committee.
The Commission also found that Sanctuary had wrongly told residents it had sent Ms de Bruyn an email about missed bin duties that she had never actually received. The Commission also found that committee minutes recording that Ms de Bruyn had neglected a garden area were unreasonable. Further findings included a committee member shouting at Ms de Bruyn’s contractor within her sight over a mowing dispute.
The Commission ruled that this behaviour amounted to workplace building. It made a series of orders addressing the committee’s future conduct. This included for the chairman to cease any monitoring of Ms de Bruyn. The committee was also ordered to circulate an apology to all owners.
The Fair Work Commission made pointed findings about correspondence sent to Sanctuary by Mr de Bruyn on his wife’s behalf. It found this correspondence appeared to have been written using AI. The Commission noted that it was “lengthy, wide-ranging, replete with generalisations, repetitive, and often couched in accusatory language.”
The Commission found it “unsurprising” that the Sanctuary committee eventually chose not to engage with Mr de Bruyn in response. However, it found this response was itself problematic, noting the committee’s approach of choosing to “disengage and dig in” meant it failed to meaningfully engage with concerns raised on Ms de Bruyn’s behalf.
Alongside orders addressing the committee’s future conduct, the Commission also ordered all future communication between the parties “to be accurate and respectful.” And it specifically directed that “there is to be no use of artificial intelligence tools in the preparation of correspondence between the parties.” The order further required that communications be kept brief and to the point.
AI tools can generate convincing but entirely fabricated legal citations, case summaries and even statutory clauses that simply don’t exist. Relying on this material without independently verifying it against real sources can seriously damage your credibility before the Fair Work Commission.
In Sarah Byrne v CXC Corporate Services Pty Ltd [2026], the Fair Work Commission found the worker had relied on AI to prepare her submissions despite an earlier warning. This included citing one case that did not exist, and two others that had no relevance whatsoever to her claim. The Commission found her declaration confirming she had checked the material “proved to be misleading,” and this significantly undermined her credibility throughout the hearing.
There is no outright ban on using AI to help draft or organise a Fair Work claim generally. However, the Fair Work Commission has shown it is willing to restrict or ban AI use where it results in unreliable, excessive or unhelpful correspondence.
In Mrs Deidre De Bruyn [2026], the Commission banned the use of AI tools entirely in future correspondence between a caretaker and her employer. This came after finding her AI-generated complaints were “lengthy, wide-ranging” and “accusatory,” and had likely worsened the very workplace dispute she was trying to resolve.
Have you been dismissed and are unsure of your rights?
If you have been dismissed and are unsure where you stand, it is far more beneficial to get proper advice from real people who understand the Fair Work system, rather than relying on AI-generated research. Courts and tribunals have increasingly found AI-generated legal material to be unreliable and, in some cases, entirely fabricated, and this can seriously damage an otherwise strong claim.
A Whole New Approach is an employee-only firm, meaning we only ever act for workers, never employers. Led by principal Gary Pinchen and backed by more than 40 years of combined experience, we help employees across Australia navigate unfair dismissal, general protections, sexual harassment and workplace bullying claims.
Your first consultation is free, and all consultations remain private and confidential. We work on a no win, no fee basis. Call us today on 1800 333 666.
