ChatGPT can't cite the case. FairWork Mate AI can. Five real questions where it shows.

If you ask ChatGPT, Claude, or Gemini a specific Australian workplace question — “am I owed redundancy after 18 months in a small business?” — you'll get a confident, plausible answer that's generally correct but isn't grounded on any specific Australian case or section number. It's a synthesis of training data that ended months or years ago.

FairWork Mate AI does it differently. Every answer that touches workplace law pulls from a live database of 260+ Fair Work Commission and Federal Court decisions, refreshed daily, and cites the relevant case by name with a link to our summary. You can verify the source. The general AIs can't even tell you which case to look at, let alone whether it's still good law.

This article walks through five real questions and shows what each tool says, side by side. Try the same questions yourself at /advisor — free, no sign-up.

ChatGPT (May 2026): Gives a generally correct overview citing the National Employment Standards. Mentions the small business exemption. Doesn't cite a specific case or distinguish between employees-on-the-day and the recent Mondelez-style accrual disputes. Confidence: high. Verifiability: zero.

Lawpath chatbot: Refers you to a paid lawyer consultation. No specific answer.

FairWork Mate AI: Calls our redundancy calculator with your inputs (12 employees, 18 months) → returns “Yes — your employer is not exempt because they have 12 employees (the small-business threshold is fewer than 15 under s.121 of the Fair Work Act). For 18 months you're owed 4 weeks' redundancy pay under s.119.” Then cites a recent FWC decision interpreting the “regular and systematic” test for casual conversion eligibility, with a link to /cases/{slug}.

The ChatGPT answer isn't wrong. It just doesn't prove anything. You can't take it to a tribunal.

ChatGPT: Mentions reasonable refusal under the FW Act. Doesn't cite a section. Doesn't mention the right-to-disconnect provisions that came into force August 2024 (or any post-cutoff developments).

FairWork Mate AI: Cites s.62 of the Fair Work Act on reasonable additional hours, references the right-to-disconnect amendment (s.333M), and pulls a recent unfair dismissal case where an employee was dismissed for refusing additional hours and won at the FWC — with a link to that decision's summary.

That's a 2024 case. ChatGPT's training data was probably cut off before it. Even if it included it, it can't prove which case it's relying on.

ChatGPT: Generally correct — pay can't be reduced unilaterally below award rates, and changes to existing contracts need agreement. Cites no case.

FairWork Mate AI: Same correct framing, but cites a recent FWC general protections case where an employer reduced pay during a slowdown without consent and was found to have breached s.323 of the Fair Work Act. Links to the decision summary. The user knows exactly what precedent supports their position.

This is constructive dismissal territory. ChatGPT will tell you that. FairWork Mate AI tells you the same and pulls the most relevant recent constructive-dismissal case from the FWC, with the test the Commission applied and how it ruled. Sometimes the case is on the user's side; sometimes it isn't. Either way, they leave knowing the actual precedent.

This is where the calculator tools earn their keep. ChatGPT will give you a number that's wrong by either a few cents or a few dollars depending on when its training data cut off and which award version it remembers.

FairWork Mate AI calls a calculator that reads from the live award rates table (updated 1 July annually after the Annual Wage Review) and returns the exact figure for 19-year-olds with the 25% casual loading on the current Restaurant Industry Award MA000119. With a citation to the award document. Same answer, today and on 1 July 2027.

For a worker checking whether their employer is doing the right thing, the difference is the gap between “here's a confident answer” and “here's a confident answer with the case I'm relying on, the section number, and the current rate.” The first one feels like advice. The second one is something you can take to your boss, your union, or the Fair Work Ombudsman.

For an employer, the difference is between getting general guidance and getting guidance you can defend in front of the FWC. We have business customers using FairWork Mate AI specifically because the citations let them show due diligence.

Open /advisor and ask the same five questions. Free, no sign-up, 2 questions a day. If you're working through a single workplace issue, the $4.99 Day Pass gets you 25 questions in 24 hours with full case-law citations on each one.

If you want to dig deeper into the methodology, see the companion piece: How FairWork Mate AI stays current with FWC and Federal Court decisions.