How New Zealand is Tackling AI and Likeness

The New Zealand Government announced its first-ever artificial intelligence (AI) strategy in early July, with the aim of positioning the nation as a “sophisticated adopter” rather than a contributor to new foundational models in the space. The strategy outlines a “light-touch and principles-based” approach, harnessing existing legislation and regulations, rather than introducing new ones. It contains a number of practical and proactive tips for businesses.

IPH’s New Zealand-based member firm AJ Park’s summary of this guidance is available here.

As in many other nations, generative AI practices have permeated public life in New Zealand. A notable example is the recent marketing strategy of Telecommunications company Skinny, which licensed the image and likeness of one of its long-term customers and competition winners, Liz. Under contract with Skinny, ‘Liz’ has become an AI-generated mascot for the company, which can now generate images and videos featuring her likeness in advertising and marketing materials. As practices like this emerge across New Zealand, is the government’s strategy well timed, and will it be enough to ensure AI use is regulated to the necessary extent, particularly when it comes to the use of individual likeness?

Personality rights, or lack thereof

A common concern surrounding the rise in easily accessible generative AI tools is its ability to recreate individual likeness, a practice that impacts both high profile individuals such as actors and sportspersons, as well as ordinary people. An additional challenge for this light-touch approach to AI regulation is the already limited existing regulations surrounding personality rights.

“In New Zealand, we’ve got very limited personality rights (as they are referred to in some countries). There’s really no dedicated law to protect someone’s rights to their likeness or personality. At present, any legal issues surrounding likeness lead us to consider established general laws, such as the Fair Trading Act, tort of passing off, privacy, or harmful digital communications,” says AJ Park Principal and Litigation Practice Group Leader, Thomas Huthwaite.

“The problem with relying on these established laws is you have to fit a round peg into a square hole. For example, when relying on the Fair Trading Act or tort of passing off, the threshold for goodwill is high. Essentially, you have to show that your likeness is well-known enough that is someone was to use it, the public would be misled or deceived into thinking they had your endorsement or approval. You’re having to rely on heavy evidential rights, which isn’t practice for the average person – it only helps famous people. Sometimes, it’s going to be a matter of getting creative with the laws available to you – and using contracts might be one way of doing that.”

Using contracts

With the rise of generative AI, concerns have risen around organisations having the contractual right to images of established public figures, particularly in the entertainment industry.

“This use of likeness is more common now, particularly with the availability of generative AI, which played a role in the recent strikes in Hollywood around contracts for actors and studios, which essentially enabled them to effectively replicate their likeness and not pay the original actor for that extended use,” says Huthwaite.

In contrast, the aforementioned case of Liz enabling Skinny to use her identity came with full awareness of generative AI used. Skinny committed to protecting the reputation of their customer turned AI mascot with a clearly laid out contract.

“In the case of Skinny, they’ve gone heavily into contractual terms governing how they could use the likeness of their customer, Liz. She allowed them to scan her and use her movement, facial likeness, and mannerisms, so they could treat her as their model, or mascot, and given the same kinds of rights that a modelling contract might have,” says Huthwaite.

“In addition, there are things like defamation legislation, which could grant additional or inherent protections for cases like that of Liz,” adds AJ Park Litigation Principal Sean Brogan. “The individual would also be protected from anything that would be illegal. Defamation would protect them from being used in a context that is unfavourable or illicit. It’s important to note though that defamation can be difficult to enforce – you need to have a reputation for it to be damaged. If no one knows who you are, does anyone actually care?” 

Critics of New Zealand’s light-touch approach have raised concerns that it is not adequate in ensuring average people are protected when it comes to genAI and use of their likeness. At the other end of the spectrum, Denmark recently rolled out copyright law entitling individuals to legal ownership of their physical likeness, covering face, body, and voice, ensuring inherent protections if these are used, or misused.

However, New Zealand has implemented regulations relating specifically to online misconduct – the Harmful Digital Communications Act, which may apply to an individual who has alleged they have suffered or will suffer harm as a result of a digital communication. Its broad positioning means it may be well-suited to cases of generative AI misuse that are likely to cause harm.

A flexible approach as AI evolves

Huthwaite and Brogan both share the opinion that combining a flexible approach with current legislation is adequate for the nation in its current state, as AI applications continue to evolve and change.

“New Zealand’s approach is fairly light touch compared to quite a few other jurisdictions which have AI specific legislation. The New Zealand Government has decided that the current framework of legislation is acceptable and that the issues around use of AI can be dealt with under what’s currently in place, such as Privacy legislation, and the Harmful Digital Communications Act, and Defamation Legislation. IP Legislation also assists in the way of copyright and trade marking an image or name,” says Brogan.

“Some other legislations have been quite prescriptive in how they have rolled out AI legislation, including applying classifications to organisations or AI tools. However, I think of this as an evolving piece – a lot of changes are coming, and legislation will need to adapt. New Zealand’s approach is a good one to ensure it can remain flexible and adapt through further changes,” Brogan concludes.