AI-Washing Just Became a $100 Million Problem in Australia

Australia doesn't have an AI Act. It doesn't need one to fine you $100 million.

Since 28 March 2026, the maximum penalty for misleading conduct under the Australian Consumer Law has doubled — from $50 million to $100 million per contravention. The actual penalty is the greatest of that $100 million, three times the benefit obtained, or 30% of adjusted turnover. The Treasury Laws Amendment (Doubling Penalties for ACCC Enforcement) Act 2026 applies to every misleading claim a business makes. Including claims about artificial intelligence.

AI-washing is the practice of overstating what AI actually does in a product or service. Calling a rules engine “AI-powered.” Marketing a keyword-matching chatbot as an “intelligent assistant.” Claiming a scheduling tool uses “machine learning” when it runs fixed logic. Corrs Chambers Westgarth, in their 2026 analysis of enforcement risks, flags three specific patterns: claiming services are AI-powered when they rely on manual processes, overstating how operational a solution actually is, and hiding the role of human intervention.

The pressure to dress things up is real. A 2026 Carta survey of 500 Australian startup executives found that four in five believe there is an AI bubble — yet most feel compelled to integrate AI into their positioning to maintain credibility with investors and customers. More than 40% said AI has already changed how they pitch when raising capital. That pressure flows downstream to every vendor selling software to your business.

In October 2025, the ACCC took Microsoft to Federal Court for allegedly misleading 2.7 million Australian subscribers. The allegation: when Microsoft integrated its AI assistant Copilot into Microsoft 365, it told existing customers they had two choices — accept Copilot with a 45% price increase ($109 to $159 per year for Personal plans), or cancel. A third option — Classic plans at the original price, without Copilot — existed but was hidden until a subscriber initiated cancellation.

ACCC Chair Gina Cass-Gottlieb said Microsoft “deliberately omitted reference to the Classic plans” to drive enrolment in higher-priced AI plans. The case is ongoing.

This isn’t AI-washing in the “our product isn’t really AI” sense. It’s something adjacent: using AI integration as cover for a price hike while burying the non-AI alternative. The principle is the same. If your communications about AI mislead customers — whether about what the AI does or what it costs — the ACL applies. And the ACCC has said it “will seek the highest penalties appropriate in any cases we bring to the courts.”

If you’re buying AI tools, ask harder questions. For trades businesses evaluating scheduling, quoting, or dispatch tools that claim AI capabilities, the test is straightforward: ask the vendor what data the model was trained on, whether it improves over time with your data, and what it does that a standard rules engine couldn’t. If they can’t answer clearly, you’re likely paying an AI premium for non-AI software. That’s a cost intelligence problem hiding in plain sight.

If you’re selling services and you’ve added “AI” to your website or marketing, the same rules apply in reverse. The ACL doesn’t distinguish between a tech giant and a 10-person accounting firm. If you claim your advisory service uses “AI-powered analysis” and it’s actually a spreadsheet with conditional formatting, that’s potentially misleading conduct.

We’ve written before about the gap between AI adoption and AI value among Australian SMEs — only 5% report getting real value from their tools, per Deloitte. AI-washing makes that gap worse. Businesses pay premium prices for capability that doesn’t exist, get poor results, and conclude that “AI doesn’t work for businesses like ours.” The technology gets blamed for the vendor’s dishonesty.

First: what specifically does the AI do that couldn’t be done with standard automation? If the vendor can’t articulate the difference, the “AI” label is marketing. Second: does the system learn from your data over time, or does it run the same logic for every customer? True AI adapts. Branded automation doesn’t. Third: can you see the output with AI disabled? Some platforms let you toggle AI features off. If the tool works identically without them, you have your answer.

These aren’t gotcha questions. A good vendor will welcome them. And in a market where penalties for misleading claims just doubled, the vendors who can answer clearly are the ones worth paying.

Key takeaways

Not sure if your AI tools are delivering what they promised?

A short conversation can help you audit your current AI spend, separate real capability from vendor marketing, and identify where the actual value sits. Book a call.