Australia now requires AI use to be disclosed in workplace cases—what changes?
From 20 October, anyone using generative AI to prepare a document for Australia's Fair Work Commission must say so, explain how it was used and verify the result. Witness statements face an extra safeguard: they must remain the person's own knowledge, words and truthful account.
The short version
- What happened
- The Fair Work Commission says its requirements apply to every Commission matter from 20 October 2026 whenever generative AI is used to create, write, edit or otherwise prepare a document.
- Who it affects
- A worker or small employer can still use generative AI to help prepare a case, but must be transparent about that help and personally verify the result before filing. A representative cannot treat a plausible-looking citation as checked law.
- What remains unproven
- The sources are the Commission's own implementation notice and guidance; this is direct policy evidence, not an independent evaluation.
Keep in mind
The sources are the Commission's own implementation notice and guidance; this is direct policy evidence, not an independent evaluation.
No denominator is published for AI-assisted filings, disclosed use, detected errors or fabricated legal authorities.
The Commission has not yet reported compliance, enforcement, processing-time or access-to-justice outcomes under the new requirements.
The August guidance predates implementation; this report preserves that original date and uses 20 October only for the effective change.
Source analysis · 3 linked sources. Primary analysed source date: 20 October 2026. Earliest linked supporting source: 24 August 2026.
Read the full report · 4 min The evidence, methods, implications and limitations
The obligation follows the document, not the brand of tool
Australia's national workplace-relations tribunal has moved from general advice to an operational rule. The Fair Work Commission says the requirements cover applications, submissions, witness statements and any other document prepared for a case. Its public guidance names ChatGPT, Claude, Copilot and Gemini as examples, but the trigger is the use of generative AI—not a particular product. A person must tell the Commission that AI was used and explain what it did.
That matters most for self-represented workers and employers, who may use a chatbot to organise facts or draft formal language. The rule does not prohibit that help. It makes the human filer responsible for checking accuracy, relevance, real evidence and genuine legal authorities before the document enters a case. The Commission also warns that a chatbot is not a substitute for legal advice.[1][2]
Witness evidence cannot be outsourced to generated prose
The strongest safeguard applies to a witness statement or declaration. The person must check that the account is based on their own knowledge, uses their own words and is true to the best of their knowledge. That creates a clear boundary: generative AI may assist with preparation, but it cannot supply remembered events, certainty or testimony that the witness does not possess.
Legal practitioners and other professional representatives carry an additional practical duty. When a document prepared with generative AI mentions case law, they must include hyperlinks to those authorities. The requirement is designed to make a cited decision easier to inspect, but a link alone does not establish that the case is real, relevant or correctly interpreted; the Commission still requires the document to be checked.[1][2]
The distinction that matters
What an AI-assisted filing now requires
The Commission places three linked checks between a generative-AI draft and the official case record.
Disclose
State that GenAI was used and explain how
The Commission can see where AI assisted the document
Verify
Check facts, relevance, evidence and legal authorities
The filer remains responsible for the submitted material
Own the evidence
Keep witness accounts truthful and in the witness's own words
Generated prose cannot replace personal knowledge
This is a policy change, not evidence of safer case preparation
The Commission released the final guidance and supporting resources on 24 August, then brought the requirements into force on 20 October. The implementation notice confirms the obligations, but it does not report how often parties use AI, how many filings contain fabricated material, whether disclosure is complete or how the Commission will measure compliance. No outcome data yet show that the policy reduces errors or improves access for people without a lawyer.
The useful test will come from case administration: whether disclosures are understandable, whether fabricated authorities fall, whether witnesses remain confident that statements reflect their own account, and whether checking creates manageable work rather than delay. Until those measures exist, the evidence supports describing a new accountability process—not claiming that AI-assisted workplace cases have become safer.[1][2][3]