By Ross Heron | CEO, Australian Payroll Association
Somewhere in Australia tonight, an employee will photograph a payslip, drop it into an AI chatbot and ask a simple question. Am I being paid correctly? Within seconds they will have an answer referencing their award, classification, penalty rates and superannuation. It may not be perfect. It will be confident enough to prompt the next question: what do I do about it?
For most of our profession's history, the ability to interrogate a pay outcome sat with the employer. Checking a year of payslips against a modern award was work for a consultant or a very patient lawyer. That asymmetry has collapsed in about eighteen months.
The detection gap has flipped
The regulator's own numbers tell the story. Anonymous tip-offs to the Fair Work Ombudsman jumped roughly 50 per cent in 2024-25 to more than 25,000, a third from workers under 24. Employment lawyers are seeing what sits behind that surge: employees feeding payslips and timesheets into AI and asking directly whether they have been underpaid. Meanwhile, recent research found that while most businesses believe their payroll is accurate, more than four in ten have found an error at some point. Confidence and accuracy are not the same thing. Employees armed with AI are now testing the gap, at scale, for free and without telling anyone.
A global shift, sharpest here
This is not a local quirk. Regulators worldwide are putting pay data into employees' hands just as AI gives them the means to analyse it. The EU Pay Transparency Directive, in force since June 2026, grants workers the right to their own pay data and comparable averages and bans pay secrecy clauses. Similar rules are spreading through the UK, Brazil, Japan and parts of the US. Layer on the fact that AI use is now routine for a large share of the workforce and the conclusion writes itself: many of your people have both a reason to check their pay and a tool that will do it in under a minute.
What makes Australia the sharp end is consequences. Intentional underpayment is now a criminal offence. The Fair Work Ombudsman recovered $358 million in 2024-25 and secured a record $23.7 million in penalties. Elsewhere, an employee's AI-assisted discovery leads to an awkward conversation. Here it can lead to remediation, a regulator and at worst, a courtroom.
One irony worth noting: over half of employees are uncomfortable with AI processing their pay, yet entirely comfortable using their own AI to audit it. Trust follows control.
Find it before they do
The right response is not fear. It is making sure the first person to find an error in your payroll is you. That means continuous assurance rather than annual sampling, classification and superannuation checks as standing disciplines and clear, defensible evidence of accuracy ready for the day someone asks. Because someone will ask and increasingly they will arrive with their analysis already done.
At the start of this year I argued the defining payroll trend of 2026 would be evidence of accuracy. Nine months on, I will sharpen that. The demand is no longer coming only from regulators and boards. It is coming from the people you pay. Your employees have joined your audit function. The only question is whether you find out what they find out before they do.
In the coming weeks we will step behind the word compliance itself and look at the four pillars you need to be able to tick before you can say, hand on heart, that you are compliant.