Claims about AI in a product, and the consumer law

Section 18(1) is short and wide:

The words “or is likely to mislead or deceive” matter here. The section does not wait for a customer to be misled; conduct likely to mislead is enough. Section 18(2) adds that nothing in Part 3-1, the Part on unfair practices, limits subsection (1) by implication.

Section 29(1) opens with these words:

Fourteen paragraphs follow, (a) to (n). Four of them are quoted here:

A line such as “our assistant answers every question accurately” describes how a product performs, and paragraph (g) speaks of “sponsorship, approval, performance characteristics, accessories, uses or benefits”. Note 1 to s 29(1) says a pecuniary penalty may be imposed for a contravention. (The example is an illustration made up for this guide, not a case.)

The ACCC has named AI-written reviews as a concern. Its Chair said in December 2025: “AI may also be used to generate and disseminate large volumes of fake reviews. These reviews may be seen as more credible and persuasive by consumers, and be increasingly more difficult to detect.”

For testimonials the Act shifts the starting point. Under s 29(2), in a proceeding about a representation of the kind in paragraph (e) or (f), “the representation is taken to be misleading unless evidence is adduced to the contrary.” Section 29(3) then says that rule does not make a representation not misleading merely because such evidence is adduced, and does not put on anyone an onus of proving that it is not misleading.

On 17 December 2025 the ACCC released an industry snapshot on AI. Among the consumer risks it listed was the use of AI to facilitate false representations. The Chair, Gina Cass-Gottlieb, said: “We are already seeing instances where generative AI is being used to facilitate false representations about the performance or characteristics of a product or service”. She gave two examples:

• “Ghost websites, which misrepresent themselves as local businesses, often use generative AI images to build a sense of credibility.”
• “Online product listings may use generative AI to make products appear more sophisticated, or of a higher quality, than they actually are.”

Those are the regulator’s observations in a media release, not findings by a court. This guide cites no court case or ACCC enforcement action about a business overstating the AI in its own product, because none appeared in the official sources it was written from, read on 9 October 2026.

The National AI Centre makes the same point from the other side: misleading statements made through AI-driven advertising or promotion could breach consumer law.

The ACCC says a claim must be based on reasonable grounds. For a representation about a future matter, the Act says so itself. Section 4(1) reads: “If: (a) a person makes a representation with respect to any future matter (including the doing of, or the refusing to do, any act); and (b) the person does not have reasonable grounds for making the representation; the representation is taken, for the purposes of this Schedule, to be misleading.” The national guidance gives a business a way to know what its AI can actually do. Among the next steps under its fourth practice, it asks an organisation to identify, document and communicate what each AI system can and cannot do, so that people neither lean on it too heavily nor misuse it. Its questions for suppliers look for written limitations and known failure points, and for evidence of testing, not just assurances.

The guidance also asks that people know when they are dealing with AI: a customer getting information from a chatbot, it says, should know it is not a human specialist. The supplier questions behind those records are in buying an AI product, and the practice itself in the six essential practices.