AI-assisted Fair Work Commission claims: what employers need to know

AI is hailed as a driver of efficiencies and innovation for employers. However, in litigious matters with employees and ex-employees, AI is proving the opposite. Applicants who use using AI in Fair Work Commission (FWC) matters, ostensibly a ‘no costs’ jurisdiction, are generating more expensive and time-consuming claims for employers to manage. 

In this Insight, we discuss the trend of AI-use in employment-related claims, the risks that arise for employers, and practical measures for employers when tackling AI-inspired claims.

AI and the law: a snapshot

Publicly available AI models can assist parties with drafting documents, conducting research and summarising legal principles. The long-term potential to increase access to justice and reduce litigation costs is clear.¹ However, there are pitfalls which are equally apparent, including ‘hallucination’ of case law, generation of factually incorrect evidence and repeated raising of irrelevant or legally flawed considerations.²

The unique vulnerability of FWC disputes to AI misuse                     

Issues stemming from the misuse of AI in legal proceedings are not unique to the FWC,³ but its distinct legislative structure and role in the resolution of workplace disputes make it especially vulnerable to abuses of AI. Key drivers include: 

• The FWC is not bound by rules of evidence in the same way that courts are. The FWC adopts less formal evidentiary requirements and its statutory mandate includes conducting proceedings in a manner that is "quick, informal and avoids unnecessary technicalities”. An inevitable result is greater ability for employees or ex-employees to institute unmeritorious claims.   
   

• Costs in FWC proceedings (including those commenced in the courts) are only awarded in limited circumstances and generally only for vexatious or unreasonable conduct. This is a stark contrast to the standard approach of a court in civil litigation where costs ordinarily ‘follow the event’ and a successful party can usually recover a portion of their legal costs. The ‘no costs’ regime in the FWC facilitates access to justice but also has the effect of incentivising continued agitation of spurious claims.  
   

• There is no entitlement to legal representation in FWC proceedings. Representation by a lawyer or paid agent requires consent of the other parties or permission from the FWC. This means many employees and ex-employees are not legally represented in FWC disputes, and so presumably don’t obtain advice about the relative merits of a claim. 

The FWC’s response to AI: a timeline

Two recent decisions featuring AI use in FWC disputes 

Two decisions delivered in August 2026 illustrate the opportunities and challenges of AI use in FWC matters. 

• In Baker v Macquarie University (Baker)⁴ the employee, a casual academic, succeeded in his application for conversion to part-time employment. Mr Baker attributed his success to the use of paid AI agents including ChatGPT Pro. Mr Baker deployed AI agents to draft documents, inquire into case law and pre-empt his employer’s likely counterarguments. Mr Baker even said it was an AI tool that first alerted him to the possibility of bringing his claim. 
   

• In Khan v Aldi Pty Ltd (Khan)⁵ an employee alleging unfair dismissal relied frequently and heavily on AI in preparing documents throughout the proceedings. Mr Khan received two communications from FWC Deputy President Easton prior to the hearing, explaining that his claim was unlikely to succeed on his own evidence. Despite this, Mr Khan's AI-generated submissions continued to address the wrong legal question and failed to engage with the concerns raised. In a step described as “very rare” by Deputy President Easton, a costs order was made against Mr Khan due to, among other things, his unreasonable ignorance of concerns raised with him.⁶

These two decisions illustrate the spectrum of AI use in FWC disputes. In Baker, a sophisticated user deployed AI tools to focus a narrow legal argument and succeeded in his claim. In Khan, what appeared to be unfettered use by an applicant of AI led that applicant to pursue a doomed claim in the face of clear warnings. But the common thread is cost for employers. Even where the employer prevailed in Khan, the evidence of the employer was that it had incurred significantly more in legal fees than the $1,230 costs order recovered. 

The inconveniences to employers

1. It is almost inevitable that employees or ex-employees using AI in FWC proceedings will lead to more voluminous claims for employers to manage and in many cases, with causes of action that are legally spurious. 
    

2. Research commissioned by the FWC identifies a spectrum of AI engagement among applicants.⁷ At one end, 'GenAI-assisted' users treat AI as one input among many, cross-referencing outputs and exercising independent judgment. At the other, 'GenAI-dependent' users accept AI outputs at face value, treating the tool as a primary source of guidance on case strength and likely outcomes. It is these GenAI-dependent users who are more likely to present with inflated expectations, resistance to settlement, increased case management needs and claims with weaker merit. These risks arise from how AI is used, not simply whether it is used, which is a distinction with direct implications for employers assessing the likely trajectory of a dispute.

AI can overstate to an employee the merits of their claim. As observed by President Hatcher, it is “common for AI tools to give a sheen of legal plausibility to cases which, on analysis, do not have reasonable prospects of success”.⁸ The AI Research Report further found AI's 'sycophantic' tendency (i.e. its design to tell users what they want to hear) inflates applicants' expectations and can make them resistant to settlement.⁹ Conciliators reported that applicants using AI arrived with “entrenched views on entitlements” that proved resistant to reality-testing, and parties quoted remedy estimates generated by AI representing the upper end of or beyond realistic ranges.  

The practical effect for employers is more claims with weak prospects of success and greater engagement with AI-dependent applicants who are less likely to accept reasonable settlement offers including by genuinely compromising on their own position, prolonging disputes and increasing costs. 

Misguided or erroneous submissions generated by AI may unnecessarily broaden the issues in dispute but cannot be ignored. In Re De Bruyn,¹⁰an employer’s choice to ‘disengage and dig in’ resulted in a failure to properly engage with matters properly raised and which ultimately weakened  their defence. Khan also demonstrates that addressing claims devoid of a legal basis can still come at significant expense. 

Practical measures to combat claims 

We recommend employers take the following steps.

Maintain diligent recordkeeping for workplace decisions 

This removes room for ambiguity. General Protections and Unfair Dismissal claims are among those most likely to be made with AI assistance. Defending both often involves an employer proving the reasons for which a particular decision was made, while defending the latter will also typically require evidence of procedures undertaken in connection with a dismissal. 

Uncertainty surrounding those reasons and procedures generates opportunities for speculative submissions and misleading or incomplete ‘evidence’ about what allegedly occurred. Those opportunities may be taken up by AI, resulting in wider, lengthier and more costly claims. Employers can mitigate these effects through robust recordkeeping.

Ensure employees understand their confidentiality obligations in respect of AI-use 

Many employees are subject to confidentiality obligations under their contract of employment.  Indiscriminate entry of information into public AI tools may amount to breaches of those obligations. Awareness of these obligations and an openness to enforcing them where appropriate may now represent a valuable strategic tool for employers than ever before.   

Take a proactive response to apparent any misuse of AI during proceedings 

If an employer suspects improper use of AI by an employee resulting in incorrect or incoherent legal arguments or persistent pursuit of a deficient case, those matters and the resultant cost risks should be raised with the employee and/or the FWC promptly.

Stay alert to developments within a rapidly changing area 

In June 2026, the Government introduced the Building Cooperative Workplaces Bill to streamline the FWC's procedures, in part to address the AI-driven surge in caseload. Together with amendments to the FWC Rules that accommodate the use of AI in the FWC, the AI Guidance Note which will enter force in October 2026 and the Government's broader regulatory agenda,¹¹ there is a concerted legislative response to the challenges outlined in this Insight that is rapidly developing.  

Employers should remain alert to developments and to changes in obligations and opportunities which flow from reforms addressing use of AI in employment-related claims. 

^{[1] See Rogers v McDonald’s Australia Ltd (AI-use) https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/FCA/2026/1264.html FCA 1264 [1]–[15] for a discussion of the use of AI in the legal field.}

^{[2] See, for example, Byrne v CXC Corporate Services Pty Ltd [2026] FWC 2751 [15]–[23]; Tischler v Avada Traffic Pty Ltd [2026] FWCFB 174 [3]; Hoverd v M & J D Pty. Ltd [2026] FWC 1013 [3]–[6].}

^{[3] See, e.g. Shahin v Industrial Relations Secretary on behalf of Multicultural NSW (No.2) https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/nsw/NSWIRComm/2026/11.html NSWIRComm 11; Barron v BT Funds Management Ltd (Leave to File) https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/nsw/NSWIRComm/2026/11.html FCA 1220.} 

^{[4] Baker v Macquarie University https://www.austlii.edu.au/cgi-bin/viewtoc/au/cases/cth/FWC/2026/3054/ FWC 3054.}

^{[5] Khan v Aldi Pty Ltd https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/FWC/2026/3144.html FWC 3144.  Note, an appeal has been lodged against the Khan decision pursuant to section 604 of the Fair Work Act 2009 (Cth).}

^{[6] Note, the costs order was made against Mr Khan under section 400A of the Fair Work Act 2009 (Cth), which is a provision specific to unfair dismissal proceedings that allows the FWC to order costs where a party's unreasonable act or omission caused the other party to incur costs. This is distinct from the FWC's general costs power under section 611.}

^{[7] Pivot Management Consulting, Fair Work Commission: GenAI User Research (Report, Fair Work Commission, 18 August 2026).}

^{[8] Adam Hatcher, 'Presentation to the Victorian Bar Association' (Speech, 18 February 2026) 6.}

^{[9] Pivot Management Consulting, Fair Work Commission: GenAI User Research (Report, Fair Work Commission, 18 August 2026) 8–9; see also Adam Hatcher, 'Presentation to the Australian Industry Group PIR Conference' (Speech, Canberra, 8 September 2026) 10.}

^{[10] Re De Bruyn https://www.austlii.edu.au/cgi-bin/viewdoc/au/cases/cth/FWC/2026/2887.html?context=1 FWC 2887.}

^{[11] On 15 July 2026, the Prime Minister announced a decisive shift in Australia's approach to AI regulation, including mandatory national standards, which was followed by a subsequent announcement of the establishment of a Fair AI Taskforce.}