FWC flags new expectations on GenAI use – with costs consequences to match

The Fair Work Commission (FWC) has published a detailed draft guidance note on the use of generative AI (GenAI) in proceedings. The timing, landing the day before Hoverd, is no accident. The guidance sets out mandatory disclosure, verification and declaration requirements, and puts costs squarely on the table for those who don't comply. Here's what it means for your practice.

Relevant documents: Draft Guidance Note: Use of Generative Artificial Intelligence in Commission cases (24 March 2026); President’s Statement (24 March 2026)

Disclaimer: The information provided in the articles in this section are of a general nature and should not be construed as specific advice or relied upon in lieu of appropriate professional advice. Whilst LEAP Legal Software uses commercially reasonably efforts to ensure the information in these articles are up to date at the time of publication, LEAP does not warrant their accuracy, currency or completeness and excludes all loss or damage howsoever arising (including through negligence) in connection with the information contained in these articles.

Quick takeaways

• FWC draft AI guidance note published 24 March 2026 - the day before the Hoverd case

• Draft in name only (mirrors existing Federal Court obligations)

• If you use AI in Commission documents: disclose it, verify everything yourself, AI cannot check AI

• AI-assisted witness statements must be certified by the witness as their own evidence

• Practitioners must hyperlink every case relied on in AI-assisted documents

• Non-compliance risks dismissal or a costs order

• Respondents have new grounds to press early objections and costs arguments

Why this matters for employment lawyers

On 24 March 2026, the President of the Fair Work Commission published an exposure draft Guidance Note on the use of generative artificial intelligence (GenAI), together with proposed amendments to Commission forms requiring mandatory disclosure, verification and declarations where AI is used.

The guidance is still in draft form and consultation has now closed. But practitioners should not assume that means the obligations are tentative or optional. The draft guidance is detailed, prescriptive, and closely aligned with the approach now taken by the Federal Court and with the recommendations of the Victorian Law Reform Commission (VLRC).

Critically for employment lawyers, the guidance landed one day before Deputy President Lake’s decision in Reece Hoverd v M & J D Pty Ltd [2026] FWC 1013 – a decision in which the Commission dismissed a general protections claim and expressly welcomed a costs application because the case was built on AI‑generated content that did not exist.

Taken together, the decision and the draft guidance send a clear signal that users must check the output of GenAI or risk adverse consequences, including costs.

The Commission’s stated problem: GenAI and workload

The President’s Statement is clear about why the Commission is intervening.

The Commission links a 70% increase in workload over three years to the widespread use of GenAI tools by litigants, particularly in unfair dismissal and general protections matters. It notes that AI tools can:

• encourage unmeritorious applications by overstating prospects; and

• generate material that is inaccurate, misleading or entirely fabricated, imposing unnecessary cost and time burdens on respondents and the Commission.

This framing is important. The draft guidance is a procedural response to a systemic problem affecting case management and fairness.

What the draft GenAI Guidance requires

The exposure draft Guidance Note imposes three core requirements whenever GenAI is used to prepare a document filed in the Commission.

1. Disclosure

If GenAI is used, the document must state that GenAI was used. Silence carries consequences: documents may be given less weight, disregarded, lead to adverse costs orders, or even dismissal.

2. Human verification (not by AI)

The user must personally check and confirm that:

• all factual assertions are correct and supported by real evidence;

• all cases, legislation and secondary sources exist and stand for the propositions attributed to them; and

• all quotations and extracts are accurate.

The guidance is explicit: you cannot use GenAI to verify GenAI outputs. Verification must be done against authoritative sources such as Commission decisions, AustLII and the Federal Register of Legislation.

3. Witness material safeguards

If GenAI is used at any stage in preparing a witness statement or declaration, the witness must certify that the evidence:

• is based on their own knowledge; and

• is true to the best of their knowledge.

False or misleading evidence carries criminal consequences under the Fair Work Act.

The ability to use GenAI to create evidence is a new development, the current related Supreme Court Practice Note prohibits the use of GenAI to create evidence.

Additional obligations for legal practitioners

For lawyers and paid agents, the standard is higher again. In addition to disclosure and verification, practitioners must:

• comply with professional obligations; and

• include hyperlinks to every case relied on in any AI‑assisted document filed.

Failure to do so may expose practitioners to regulatory complaint, as well as costs risk in the proceeding itself.

The Commission expressly sought comment on extending the requirement to include hyperlinks to every case to unrepresented litigants.

The proposed form changes: costs front and centre

The draft example GenAI section for Commission forms makes the costs risk explicit. It requires applicants to tick-box certify their checks and warns that failure to do so may result in dismissal or a costs order.

This is a notable development in a jurisdiction that is usually described as “no costs”.

Hoverd: decided the day after, proving the point

The timing matters.

Hoverd was decided on 25 March 2026, one day after publication of the exposure draft. In that case, the Commission found the applicant had relied on:

• contractual clauses and award provisions that did not exist; and

• persisted with those contentions even after being warned.

• The applicant admitted using AI. Deputy President Lake concluded that he had chosen to rely on AI rather than reading his own contract and invited a costs application.

The draft guidance now squarely targets exactly that mischief: fabricated authorities, invented clauses, and persistence after error is exposed.

“Draft” does not mean discretionary

Although the guidance note and declaration are formally still drafts, there are strong reasons to expect they will be adopted in substantially the same form:

• The obligations mirror the Federal Court’s Use of Generative Artificial Intelligence Practice Note, including disclosure, personal verification and practitioner responsibility.

• The approach is consistent with the VLRC’s October 2025 report, which emphasised accuracy, human oversight and costs consequences for misuse.

• The Fair Work Commission has expressly tied GenAI misuse to workload disruption and unfair cost shifting, signaling an institutional commitment to enforcement rather than education alone.

In short, this is not experimental policy. It is a warning shot.

Practical takeaways for LEAP clients

• Build verification into workflow. AI‑assisted drafting must be followed by source‑level checking, not cosmetic review.

• Treat disclosure as inevitable. The question is no longer whether to disclose AI use, but how to do so accurately.

• Costs are back on the table. Where a claim is driven by AI‑generated content with no factual or legal foundation, Hoverd shows the Commission is prepared to entertain costs applications.

• Respondents should use this. The draft guidance provides a framework to press early objections, strike‑out arguments and costs where AI content has no reasonable prospects.

Used carefully, GenAI can improve efficiency. LEAP clients have the benefit of closed GenAI products that are fully integrated into the practice management system – MatterAI and LawY. LawY enables LEAP clients to quickly and accurately conduct legal research and provides hyperlinks to source documents – for quick human verification. When clients are time or resource constrained, they can use LawY’s human verification button to check the accuracy of the prompt response.

Related reading

• When AI shortcuts cost more than they save: Hoverd v M & J D Pty Ltd [2026] FWC 1013