← Field Notes

AI Compliance & Regulation

ACCC: You're Liable for What Your AI Chatbot Promises

The ACCC warned Australian businesses they're legally liable for what their AI chatbots tell customers. What it means — and how to limit your exposure.

Does software help decide things about your customers or staff? From 10 December 2026 your privacy policy may have to say so. Six questions give a likely answer.

Take the check →

Your chatbot speaks for you — legally

Early this month the ACCC's deputy chair put a point to Australian business that most owners running a chatbot haven't thought through: if your AI assistant tells a customer something that isn't true, the law treats it as your business saying it. Not the bot's problem. Not the vendor's. Yours. And the maximum penalty for misleading a customer in Australia doubled this year — to $100 million.

In a speech in early November, reported by Law360 Australia, ACCC Deputy Chair Catriona Lowe warned that businesses remain legally responsible for misleading claims made by their AI chatbots, and for anti-competitive conduct driven by algorithms. The reasoning is simple, and it has been the ACCC's line for years: the Australian Consumer Law is technology-neutral. A misleading representation is a misleading representation — whether it comes from a salesperson, a web page, or a chatbot you switched on to save time. The warning lands ahead of unfair trading reforms the government has flagged for next year.

The stakes moved in March. Under the Treasury Laws Amendment (Doubling Penalties for ACCC Enforcement) Act 2026, from 28 March the maximum company penalty for misleading or deceptive conduct became the greater of $100 million, three times the benefit gained, or 30% of turnover during the breach — double the old $50 million ceiling. Those are worst-case numbers aimed at the largest offenders; a small business won't see $100 million. But the everyday version still bites: being forced to honour what the bot promised, refund the customers it misled, and answer an ACCC complaint.

$100M

Maximum company penalty for misleading conduct

Doubled from $50M on 28 March 2026

$2.5M

Maximum penalty for an individual

Sole traders and directors included

The "it was the bot" defence has already failed

You don't have to imagine how this plays out — a grieving traveller already tested it. After a death in the family, he asked Air Canada's website chatbot about bereavement fares. The bot told him he could book now and claim the discount back within 90 days. That wasn't the airline's policy. When he claimed, Air Canada refused — and argued, in effect, that it couldn't be held responsible for what its own chatbot said, as though the bot were a separate entity answering for itself. In February 2024 a Canadian tribunal rejected that flat, and ordered the airline to honour the price its bot had quoted.

The award was small — about CAD $812 — and the case is overseas, so it doesn't bind an Australian court. The principle is what travels: a business put an automated agent in front of customers, the agent made a representation, and the business wore it. That is exactly the logic the ACCC is now applying under the Australian Consumer Law — and here, honouring the promise is only the start of the bill.

What it means for Australian businesses

A chatbot gets a business into trouble in three ordinary ways. It invents a price, a discount or a refund policy you don't actually offer. It overstates what your product or service does. Or it gives a confident answer a customer relies on — a delivery date, an eligibility rule, a quote — that turns out to be wrong. In each case the representation is yours. A plumbing firm's after-hours booking bot that quotes a $0 call-out fee has made a promise; an accounting practice's website assistant that tells a prospect it can lodge by a date it can't meet has made one too.

This is where the quality of your AI customer service stops being a service question and becomes a legal one. We've written before about how half of Australians say AI has made customer service worse — a frustrated customer is bad for business; a misled one is a liability. The same regulator, under the same law, is also pursuing the mirror-image risk: overstating your own use of AI, which the ACCC has turned into a $100 million problem. And it sits alongside a separate online-safety duty of care now moving through Parliament for chatbots. The consumer-law exposure is the one that already applies today — no transition period, no grace window.

If you can't say exactly what your chatbot is allowed to promise a customer, that gap is worth closing before a customer finds it for you — book a call and we'll map where your bots carry risk.

What to do about it

Don't switch the bot off — scope it. Start by listing every place AI already speaks to your customers, and what it's allowed to say about price, availability, refunds and timelines. A bot that answers "what are your opening hours?" is low-risk; one that quotes jobs or makes commitments needs a tested, constrained knowledge base, not a raw model free to improvise. Give it a clean handoff to a human the moment it's out of its depth, so nothing that binds the business is left to the machine alone. And keep a short record of what the tool is and what you checked before go-live — because if you're ever asked, the question will be whether you took reasonable care to make it accurate. That's the same discipline good AI customer service needs anyway; the consumer law just raises the cost of skipping it.

Key takeaways

01The ACCC has warned Australian businesses they remain legally responsible for misleading claims their AI chatbots make to customers — the Australian Consumer Law is technology-neutral, so "the bot said it" is not a defence.
02Maximum company penalties for misleading or deceptive conduct doubled to $100 million (from $50 million) for conduct on or after 28 March 2026.
03An overseas precedent — Air Canada, 2024 — shows courts rejecting the argument that a business isn't responsible for what its chatbot tells a customer.
04The fix isn't turning the bot off. Scope what it can promise, give it a human handoff, and keep a record of what you checked before go-live.

Common questions

Is my business liable for what my AI chatbot tells customers in Australia?

Yes. The ACCC has warned that the Australian Consumer Law is technology-neutral — a misleading claim made by your chatbot binds your business the same as one made by a staff member or on your website. "It was the bot" is not a defence.

How much can a misleading claim cost under Australian Consumer Law?

For conduct on or after 28 March 2026 the maximum company penalty is the greater of $100 million, three times the benefit obtained, or 30% of adjusted turnover — double the previous $50 million ceiling. Most small-business cases are far smaller, but you can still be made to honour what the bot promised and refund affected customers.

Do I need to switch off my customer chatbot to stay safe?

No. The risk is in what the bot is allowed to promise — prices, refunds, eligibility, timelines. Scope it to answer only what it can answer accurately, add a human handoff for anything that commits the business, and keep a record of what you checked. You keep the tool without the exposure.

Sources

Assumptions & methodology
  1. ACCC Deputy Chair Catriona Lowe's warning is as reported by Law360 Australia (6 November 2026), covering a speech given the preceding Thursday; the ACCC's own transcript was not located at the time of writing, so the remarks are attributed to that report. The ACCC has named misleading claims about AI as an enforcement focus and has long pushed for unfair trading practice reforms, which the government has flagged for 2027.
  2. The maximum penalty figures are from the Treasury Laws Amendment (Doubling Penalties for ACCC Enforcement) Act 2026, which doubled the fixed corporate maximum under the Australian Consumer Law from $50 million to $100 million for conduct on or after 28 March 2026. The maximum is the greater of $100 million, three times the benefit obtained, or 30% of adjusted turnover during the breach; the individual maximum of $2.5 million was unchanged. These are statutory ceilings for the most serious conduct — realistic exposure for a small business is far lower, and the practical cost is usually being required to honour the representation, refund affected customers and respond to an ACCC complaint.
  3. Moffatt v Air Canada was decided by British Columbia's Civil Resolution Tribunal in February 2024. It is a Canadian decision, not binding in Australia, and is cited here only to illustrate how the "the chatbot is responsible, not us" argument has been treated. The tribunal ordered Air Canada to pay roughly CAD $812 (refund, interest and fees); reported figures vary slightly across outlets.
  4. Last reviewed 8 November 2026. General information, not legal advice. Confirm your obligations with a qualified adviser before relying on this. CoterieLabs is an Australian AI consultancy; we are not a law firm.
ShareLinkedInEmail

Field Notes are general commentary on AI trends for Australian businesses. They don’t constitute professional advice. Talk to your accountant, lawyer, or IT adviser before acting on anything specific to your situation.

If this applies to you

Does your chatbot make promises you'd have to keep?

Book a call. We'll map where AI already speaks to your customers, flag the answers that could bind you under consumer law, and set the guardrails — scope, human handoff, a short record — that let you run a chatbot without wearing its mistakes.

Don't miss the next one

Get each new Field Note in your inbox as it publishes — short, practical AI intelligence for business owners.