It sounds like science fiction, but it’s reality. TikTok star Khaby Lame recently struck a $975 million deal with a Hong Kong investment firm. The reason? They want to use his face. Not just for traditional adverts, but also to create an AI version of him. Welcome to the era in which your likeness has become a commodity.
From a Nespresso advert to an AI avatar
The idea that a familiar face can generate revenue is nothing new. George Clooney has been drinking Nespresso on your screen for years. But what is new is that companies no longer just want to capture your face on camera. They want to be able to recreate it. Copy it. Make it speak languages you don’t even speak yourself. And do so 24 hours a day, without you even having to be there.
Khaby Lame, the Italian TikToker who rose to fame by remaining silent in his videos, will soon be speaking via AI. In multiple languages. And livestreaming. Without being behind the camera himself. The Hong Kong-based company Rich Sparkle expects to earn over $4 billion from him within three years. The deal means that they will decide where his face appears and what projects he takes part in.
Bruce Willis as a digital twin
Khaby Lame isn’t the first. The American actor Bruce Willis had already taken part in a similar project back in 2022. A Russian telecoms company used 34,000 photos to create a digital replica of his face. That digital twin then appeared in advertising campaigns, without Willis himself having to be on set.
We’re seeing this trend closer to home too. Former footballer Ruud Gullit recently launched an AI clone of himself, specifically designed for brand collaborations. “In 5 to 10 years’ time, this will be completely normal,” he said at the launch. He’s probably right.
The sporting world is getting involved too
The trend is not limited to one-off deals with individual celebrities. In the US, the Major League Baseball Players Association, the trade union representing professional baseball players, has entered into a partnership with the technology company Genies. The aim? To create AI avatars of the biggest baseball stars.
Fans will soon be able to interact with digital versions of their favourite players all year round. Not just during matches, but via all sorts of digital channels. It is described as a shift from passive viewing to active engagement. The players themselves don’t have to do anything; their digital avatars take over.
The silent army of AI models
Alongside the big names, there is another group playing a part in this development. Professional models and actors are selling their faces, facial expressions and body language to platforms such as Synthesia and Hour One. Their faces are then turned into AI avatars that appear in corporate videos, e-learning courses and advertisements worldwide.
These models often sign up for a one-off lump-sum payment. After that, they have little or no say in how their digital likeness is used. It’s an attractive sum in the short term, but in the long run it can cause problems.
The legal framework: what exactly is the right to one’s own image?
To understand the legal implications of these deals, we need to go back to basics. In the Netherlands, the right of publicity is governed by article 21 of the Copyright Act. This right protects you against the publication of your likeness without consent, provided you have a reasonable interest in doing so. In the case of commercial use, that interest is, in short, almost always present.
But there are further legal implications. An AI clone is often a dual legal construct: both a likeness and personal data. As soon as such a digital version can be traced back to an identifiable person, the GDPR also applies. This means that there must be a valid legal basis for processing, usually consent.
In this context, a portrait is broader in scope than you might think. It’s all about recognisability. A photograph, video, illustration, 3D model or even a character in a series can be a portrait. The question is always: is the person recognisable?
It is interesting to note, incidentally, that according to case law, the voice is not subject to the right to one’s own image, but may constitute special category personal data within the meaning of the GDPR. An imitated voice is therefore subject to a different regime of protection.
It is also important to distinguish between commissioned portraits and those that are not. In the first category, there are contractual agreements in place. In the second category, a balancing of interests must be carried out: the subject’s right to privacy and their honour and reputation on the one hand, and the interest in publication on the other. The outcome of this balancing exercise determines whether publication is ‘permitted’.
And what about AI-generated faces?
This is where it gets interesting. With AI, you can generate new faces based on a large collection of photos. Technically speaking, those faces are unique. But what if one of those generated faces bears a strong resemblance to an existing person?
In 2022, the Supreme Court handed down an important ruling in the case between Max Verstappen and the delivery service Picnic. Picnic had used a lookalike of Verstappen in a commercial without his consent. The Supreme Court ruled that the use of a lookalike can also be regarded as the use of a person’s likeness.
This ruling is relevant to AI-generated faces. If an AI portrait bears a strong resemblance to an existing person and is used for commercial purposes without consent, this may constitute a breach of the right to one’s own image. It does not have to be an exact replica of the face; a resemblance sufficient for the public to recognise the person may be enough.
The downside: deepfakes without consent
We see the dark side of this technology in deepfakes. A pornographic deepfake was created of presenter Welmoed Sijtsma without her consent. Her face was superimposed onto the body of a porn actress. The video spread via websites and WhatsApp groups.
Sijtsma reported the incident to the police and produced the series Welmoed and the Sex Fakes for NPO 3, in which she set out to find the person responsible. With the help of an IT investigator, she tracked down the perpetrator. The court described the video as shocking and distasteful and ruled that the victim’s sexual autonomy and privacy had been seriously violated.
The defendant’s lawyer argued that the creation of deepfakes is not explicitly criminalised. The court did not agree. Deepfakes can also fall under the criminal law prohibition, provided that the footage appears so realistic that it is not immediately apparent that the images have been manipulated. The creator was ultimately given a suspended community service order of 120 hours.
In addition to criminal law, the person depicted may also invoke their right to privacy. In such cases, they may seek to prohibit the distribution of the image and claim damages. This illustrates an important point: not all AI-generated portraits are problematic, but this changes as soon as an AI-generated portrait portrays someone in a false light or is used commercially without consent.
It is good to know that a bill has been tabled to introduce a neighbouring right to protect against deepfakes. If it is passed, every person will have the right to prohibit or authorise the creation and distribution of deepfakes of their voice or likeness.
Virtual influencers and deception
Deepfakes involve impersonating real people. But there is another side to this story: entirely fictional people who have never existed. Earlier this year, an Instagram account belonging to a so-called American soldier (Jessica Foster) grew to over a million followers within a few months. The woman posed in army barracks, next to fighter jets and apparently even alongside world leaders. Everything about it looked real, but nothing was. The account turned out to have been generated entirely by AI and was used to redirect followers to paid platforms featuring explicit content.
This type of deception falls into a legal grey area. There is no real person whose image rights are being infringed. Yet consumers can certainly be harmed. Think of unfair commercial practices or scams. It also shows just how convincing generative AI has become. Whereas in the past you could rely on minor flaws in an image, these are becoming increasingly rare. Social media is thus entering a new phase: not only can posts be fake, but entire identities can be too.
Can you actually sell your image rights?
Now that we’ve covered the legal framework, let’s get back to the deals. When you read headlines about selling ‘your face’, that’s actually a bit misleading. Although the right of publicity is enshrined in the Copyright Act, it is not an intellectual property right. Your right of publicity is not a property right that you can simply transfer.
What Khaby Lame and others are actually doing, then, is granting a comprehensive licence. They give a company permission to use their image, often on an exclusive basis, often for a specific period, and increasingly including the right to create AI versions. However, the right to their own image remains with them.
It sounds lucrative, and it is. But there’s a significant downside. With a traditional photoshoot or commercial, you’re there in person. You see what’s happening and can make adjustments. With AI-generated content, it’s different. A company can create new images at lightning speed without you having any say in the matter. Your face could end up in contexts you would never have chosen.
Imagine this: your AI avatar is promoting a product you personally have no interest in. Or worse, your digital twin is saying things that run completely counter to your own beliefs. These are scenarios that need to be clearly defined in a contract, as it’s difficult to intervene once the damage is done. New technologies make certain things easier for the company, but more complicated for the person whose face is being used.
Who is responsible for what an AI clone says?
A question that is often overlooked: who is held liable if your AI clone says something that is unacceptable? In practice, the primary responsibility usually lies with the party deploying the AI clone, not with the person on whom the clone is based. Under traditional liability law, the company that develops, trains and operates the system is also the party that is held liable first. This applies all the more if the AI clone is self-learning and you, as the person portrayed, have no direct control over exactly what is said. Nevertheless, it is wise to iron this out contractually: specify who is responsible for training, monitoring and publication, and include an indemnity clause for any damages.
Aandachtspunten bij AI-portret licenties
For anyone considering this sort of deal, there are a few key points to bear in mind.
Firstly, the scope of the licence. What exactly can your image be used for? And just as importantly: what can’t it be used for? The more specific, the better. Also consider geographical restrictions and the type of content.
Secondly, the duration. Three years of exclusivity sounds straightforward, but what happens after that? Will the company be allowed to continue using the AI content it has already created? Or does everything have to be taken offline?
Thirdly, oversight. Are you given the chance to review the final product before it goes live? Do you have a right of veto? Arrangements like these can make the difference between a successful collaboration and a reputational nightmare.
Fourthly, the remuneration. A one-off lump sum may seem straightforward, but if your AI avatar generates income for years to come, you might want a share of that. Consider royalties or profit-sharing. Or, as in the case of Khaby Lame, a shareholding that allows you to share in the success.
Fifthly, protection against misuse. What if your AI portrait is used in a way that damages your reputation? What legal remedies do you have in that case? And who is liable?
And finally: what happens afterwards? If the partnership ends, is the company still allowed to use the AI clone? For how long? And is additional compensation payable for this? These are questions you need to answer in advance, not after the fact.
A market that is still in its infancy
We are only at the very beginning of this trend. The technology is getting better and cheaper all the time. This means that it’s not just A-list celebrities who are of interest for this sort of deal. Micro-influencers, presenters, athletes and models are also being approached.
If you’re considering this, it’s worth knowing what you’re agreeing to. Because taking your face off the internet is a lot harder than putting it there.
