Chief Censor: There is no AI loophole. What every New Zealander should know about creating harmful content
Caroline Flora, Chief Censor on June 23, 2026
Artificial intelligence has made it easier than ever to create realistic images and videos. But it has also created a dangerous misconception: that if something is AI-generated, it somehow sits outside the law. It doesn't. If AI-generated content meets the legal threshold for being objectionable, it can be banned in New Zealand.
Artificial intelligence has made it easier than ever to create realistic images and videos. But it has also created a dangerous misconception: that if something is AI-generated, it somehow sits outside the law.
It doesn't.
Under New Zealand's Films, Videos, and Publications Classification Act 1993, it doesn’t matter whether a publication was created with a pen, a camera, a computer or AI. What matters is the nature of the content itself. If AI-generated content meets the legal threshold for being objectionable, it can be banned in New Zealand, with serious criminal penalties for creating, possessing or sharing it.
AI-generated content isn't new to the Classification Office
We've been classifying computer and AI-generated publications for years. Whether something is real or not doesn’t always matter under the Classification Act.
AI-generated images and videos are becoming more realistic, easier to create and easier to share. For many of us, it’s impossible to tell what's real and what's artificially generated.
Under the Classification Act, however, this distinction isn’t relevant. Content doesn't need to be real, or even show a real person, to be objectionable.
What makes AI-generated content objectionable?
Under New Zealand law, a publication can only be classified as objectionable (illegal) if it deals with matters such as sex, horror, crime, cruelty or violence in a way that is likely to be injurious to the public good.
Some publications are deemed objectionable, including material that promotes or supports the sexual exploitation of children or sexual violence. In other cases, we must consider whether a publication deals with specific factors, such as exploiting the nudity of children or young people.
The technology used to create the content isn't the deciding factor. A sexually explicit AI-generated image of a child, for example, can be classified as objectionable even if no real child is shown. Or, if the content shows a real child, but the image of them has been ‘nudified’ and sexualised – this content can be, and has been, classified as objectionable.
Young people need to know the risks
Recent reporting has highlighted a worrying trend of students using AI to create explicit or degrading images, and our research shows that people have a mixed understanding about the law with respect to harmful content.
It's easy to dismiss this behaviour as a prank, but the impact on victims can be profound. Young people also need to understand they may be exposing themselves to serious legal consequences.
Growing up with AI tools doesn't mean there are no rules around using them. The technology may be new, but our rights and responsibilities under the law are not.
It’s our job to communicate with the public about the classification system, and I see this as basic civics education: making sure people understand what the law says so they don’t inadvertently break it. In this way, we’re supporting people to exercise their right to freedom of expression, by making informed choices about the content they choose to make and consume. We’ve made the law simpler to understanding by translating it into plain, practical guidance on our website.
We’ve also developed resources on issues like AI and harmful content, and we continue to publish research and insights to help people understand what’s happening and why it matters.
The Classification Act isn't designed to address every harm
It's important to be clear about what the Classification Act can and can't do.
The Act wasn't designed to address every harmful use of AI or every intimate image created without consent. For example, if an adult's likeness is used to create a fake intimate image, the personal violation isn't what the Classification Act assesses.
It also wasn’t designed to regulate AI tools themselves; like a pen or camera, the tools are not what is classified. The legislation focuses on the content that the tools generate. It asks whether the publication deals with specific criteria, and if it does, it can potentially be classified as objectionable. If it doesn't, other legal remedies may be more appropriate.
Where the Classification Act does apply, it is a powerful piece of legislation. It deals with the most harmful publications, irrespective of how they were created, and people need to be aware of the law before they make or share this content.
Final thoughts
AI is raising broader questions about consent, privacy and online safety that extend beyond classification law. As these technologies evolve, it's appropriate that New Zealand continues discussing whether our legal framework should evolve too.
But one thing is already clear: AI doesn't create a legal loophole. While the technology may change how harmful content is created, it doesn't change the legal standards that apply. If AI-generated content meets New Zealand's high threshold for being objectionable, it can be banned.
If schools, educators, or organisations are trying to navigate these issues, our resources are there to help. Get in touch, we’re here to support people working with rangatahi in a rapidly changing digital environment.
Further reading
Subscribe to our blog
Stay up to date with the Classification Office blog.
