AI Creative
AI in NZ Ads: What the ASA Guidance Says
18 November 2026
On Monday night RNZ's Nights asked whether businesses should be using AI in advertising, and whether the ones that do are putting their reputations at risk. It is the question a lot of NZ marketing teams are quietly asking. The short answer is yes, you can use AI in your ads, and the Advertising Standards Authority (ASA) has already published how it expects you to do it.
This is a plain reading of the ASA's Guidance Note on AI use in advertising, published in September. It is not legal advice, and the ASA says the note does not bind its Complaints or Appeal Boards. But it shows how the regulator is likely to read an AI-made ad, which is worth knowing before you commission one.
The starting point: AI is a tool, and you are responsible
The ASA applies the existing Advertising Standards Code to any ad, whether AI helped create, alter, target or distribute it. Advertisers stay responsible for their ads, including content supplied by a third party. Handing the work to an agency, a freelancer or a tool does not move that responsibility.
Two things follow from that.
Claims still need evidence. AI tools can produce inaccurate or exaggerated copy. Objective claims in AI-written product descriptions, social posts and demonstrations need checking and substantiating, exactly as they would if a person had written them.
Testimonials need care. AI can help edit, shorten, translate or present a genuine testimonial, provided the final version keeps the original meaning and you can show that it does. It should not be used to invent reviews or customer experiences that never happened. The ASA expects you to keep a record of the original feedback.
Do you have to label AI-made ads?
Not always. The ASA says it does not expect all advertising created with AI to be labelled. Disclosure is a judgement about materiality: could the AI use change how people understand the ad?
The guidance says to consider disclosure when AI is used to:
- create a realistic person who does not exist
- copy a real person's likeness or voice
- show events, experiences or outcomes that did not happen
- produce product imagery that materially differs from the actual product
- present AI content in a way people would reasonably take as a real person, event, endorsement or demonstration
It says disclosure is unlikely to be needed for AI-assisted editing or lighting fixes, drafting copy that accurately describes your product, a decorative background, or an obviously fantastical setting that nobody would take as real.
If you do label, the label should be clear, easy to notice and close to the content. The ASA's examples include "AI-generated image", "AI-generated video" and "AI-generated voice", shown on screen or in the caption.
When a label is not enough
This is the part most people miss. The ASA treats disclosure as a transparency measure, not a fix for a misleading ad. It gives two examples where a label would not save you:
- A skincare ad using an AI-generated before-and-after image that shows results the product cannot achieve. Marking it "AI-generated image" is unlikely to be enough if the overall impression is still misleading.
- Realistic AI-made customer testimonials claiming results that no real customer reported. Calling them "AI-generated testimonials" does not change the impression that real customers had those experiences.
The ASA says the overall impression a consumer takes from an ad, which it calls consumer takeout, is a key consideration for its boards. It is a good test to run on your own work: ignore the label, and ask what a person would believe after seeing the ad.
The reputation question sits outside the rules
Meeting the code does not answer what RNZ was really asking. Some customers push back on AI-made work because they see it as coming at the expense of skilled professionals, and no guidance note addresses that.
At the same time, the industry is treating AI creative as a craft in its own right. StopPress reported last week that the Aaron Awards, which are dedicated to AI creativity in advertising, have opened entries for 2027 after a first edition that drew submissions from 45 countries.
Our view on managing the reputation risk is simple:
- Keep the things that must be true real. Use the real product, the real premises and the real people. Let AI handle settings, variations and volume.
- Match the method to the brand. If your value is craft, heritage or a human story, a fully generated ad carries more risk than it does for a brand selling on price and convenience.
- Be ready to answer "did you use AI?" A straight answer, given without fuss, costs less than a customer finding out on their own.
Five questions to ask before you publish
These come straight from the ASA's guidance:
- Could people reasonably believe the AI-generated content is real?
- Could it change how they understand the product, service, person or event shown?
- Could the ad mislead through omission, implication or overall presentation?
- Is every objective claim supported by evidence?
- Would a label help prevent misunderstanding?
If the answer to any of them is yes, rework the ad and consider a label. One habit that helps: keep a short record of what was generated, what was real, and what evidence supports each claim. If a complaint ever arrives, you will want that in front of you.
The practical takeaway
For most NZ businesses, the rules are lighter than the headlines suggest. Using AI for editing, drafting, backgrounds and stylised work needs no label. The risk sits in realistic people, invented outcomes, product imagery that drifts from the real thing, and fake testimonials, and a label does not fix any of those.
The ASA's site also lists a new Advertising Standards Code from February 2027, so it is worth checking the guidance again when that lands. In the meantime, our guide to AI video production in NZ covers where AI video still falls short, and our guide to choosing an AI creative agency in NZ covers what to ask a team before you hire them.
Work with us on this
Muscle+Brain produces AI video and imagery for New Zealand businesses and brands, built around your real product and approved before it goes out. If you are planning a campaign and want a straight view on where AI fits and where it does not, we are happy to talk it through.