The $100 Million Reason to Check What Your AI Says About You
- Elaine Schillinger
- Jun 16
- 5 min read
Updated: Jun 17
Your AI is making claims on your behalf every day. Can you prove all of them?
In my consulting work I come across this regularly. A business has used AI to draft their website copy, their proposals, their client emails. The writing is clean and professional, but buried somewhere in the output is a claim the AI generated entirely on its own. It wasn't in the brief, is not accurate and in some cases, has been sitting in front of clients for weeks before anyone noticed.
And nobody checked it.
What used to be an embarrassing quality problem can now be a $100 million one.
From 28 March 2026, the Australian Competition and Consumer Commission (ACCC) doubled the maximum penalty for misleading and deceptive conduct under the Australian Consumer Law, from $50 million to $100 million per contravention. At the same time, the ACCC flagged AI-washing as a specific enforcement priority for 2026-27.
AI-washing is what happens when a business makes inaccurate claims about their products and services using AI-generated content. The ACCC is not interested in whether a human or a machine wrote the claim. If it is misleading and your name is on it, you are responsible for it.
This applies to every business using AI to produce external communication, not just those in regulated industries.
The rule is the same. What changed is everything else.
At the risk of sounding old, I spent years in marketing, from advertising and PR campaigns at Pacific Brands to managing flash memory products when digital cameras were taking off, and the one constant across all of it was this: a business must be able to prove any claim it advertises.
Every compliance training session I was required to do, every legal sign-off on a product performance claim, every piece of scrutiny that landed on a TV ad pushing the boundaries came back to the same principle:
You are responsible for what you say, and you had better be able to back it up.
The process was imperfect even then. Reebok found that out in 2014 when the Federal Court of Australia ordered them to pay $350,000 after the ACCC alleged their EasyTone shoes, promoted through packaging, swing tags and in-store materials as increasing muscle tone more than regular walking shoes, could not be substantiated. Those claims were written by humans, reviewed by humans, and printed on millions of boxes. A human made the decision to say what was on that box, and a human failed to verify it. (Source: ACCC v Reebok Australia, Federal Court, 2014.)
With AI-generated content, nobody makes that decision. The tool infers and fills gaps with plausible-sounding content that nobody briefed and nobody explicitly approved. Because AI output looks polished and authoritative, it does not trigger the same alarm bells a rough draft might. It looks like finished work, it reads like something a professional wrote, and so it goes out the door.
The risk lives in the plausible ones.
AI's obvious errors are relatively easy to catch. The harder problem is the content that is almost right, the claim that sounds like something you might have said, that a busy reviewer would nod at and move past.
A class action filed in California in 2004 against flash memory card manufacturers including Kodak, Lexar and SanDisk comes to mind. I was working in that industry at the time, and the case was a regular conversation in the market. The allegation centred on a measurement convention, decimal gigabytes, that was technically standard but created a false impression for consumers who ended up with less usable storage than the packaging implied. The manufacturers were working within a defensible industry standard. The problem was that it meant something different to the people buying the product, and the gap was invisible to anyone who was not specifically looking for it. (Source: Vroegh v Eastman Kodak Co. et al., California Superior Court, settled 2006.)
That is exactly what AI does. It generates plausible, fluent, confident content, with a gap between what is supported by your source material and what the tool produced independently, and you cannot see that gap unless you are looking for it.
Confident output is not a green light. It is a prompt to ask better questions.
So what does responsible practice actually look like?
The good news is that it does not require a new system, a compliance team or a technology audit. It requires a habit.
In my work I use a verification step called the A/B/C method. Before any AI-generated content goes to a client, a customer, or a public audience, every claim in the output gets a label:
A | DIRECTLY SUPPORTED Present in the source material you provided. Verified. It can stay. |
B | IMPLIED BUT NOT STATED The AI inferred this. Reasonable, but needs a human judgement call. |
C | NOT SUPPORTED AI generated this independently. Verify it, or cut it. |
That invented capability claim buried in a client's website copy? It was a C that looked like an A. That is the whole problem, and it is exactly what the A/B/C step is designed to catch.
Three minutes, applied consistently, every time something goes out the door with your name on it. The same rigour that used to go into a product performance claim on a swing tag, applied to what your AI just generated in thirty seconds.
Then the algorithm cases arrived, and the stakes got much higher.
In 2022, the Federal Court ordered Trivago to pay $44.7 million in penalties after the ACCC found the company's recommendation algorithm had misled consumers into believing they were seeing the cheapest hotel deals, when the algorithm was actually prioritising hotels that paid higher fees. The court found the conduct was systemic and that the design of the algorithm was intentional. (Source: ACCC v Trivago N.V., Federal Court, 2022.)
That case established something every business using AI should understand: algorithmic outputs are subject to exactly the same consumer law scrutiny as human-generated claims. The ACL does not ask whether a human or a machine produced the content. It asks whether consumers were misled, and who is responsible. The answer is always the business whose name is on the content.
The Reebok penalty was $350,000 in 2014, Trivago paid $44.7 million in 2022, and the ceiling is now $100 million. The direction of travel is not ambiguous.
The discipline has not changed. The context has.
What I took from those years in marketing was a habit of mind rather than a checklist. You treat every claim as something you might have to defend, you ask whether you can prove it before it goes anywhere, and you build that check into how you work. The discipline came from experience, from having seen what happened when the process failed.
That habit is what most businesses using AI in their external communications are missing right now. AI-generated content does not look like a draft that needs scrutiny. It looks like finished work. And that is precisely where the risk lives.
The A/B/C method is how you bring that discipline back. Label what the AI knew because you told it, label what it inferred from context, and treat anything it generated independently as something to verify before it stays. Three minutes, every time, before anything goes anywhere.
The marketing discipline has not changed. The context has, and so has the consequence of getting it wrong.
If this applies to your team
The AI in High-Stakes Communication workshop I run covers responsible AI use in practice, including the A/B/C verification method, built around real scenarios from your own environment. It has been delivered to parliamentary offices and professional services teams, and every session produces the same moment: the room gets quiet when people first see the gap between what they gave the AI and what the AI added entirely on its own.
If you want to talk about what that looks like for your organisation, I would love to have that conversation.
Comments