Skip to content 
If you woke up tomorrow to find an AI version of you selling protein powder on TikTok, what would you do? It is no longer a hypothetical. Deepfakes have become so convincing and so easy to make that the question of how to protect your face, your voice and your name has moved from the realm of science fiction into the in-tray of every IP lawyer.
Last week, the European Union Intellectual Property Office published what is being described as the first trademark application of its kind. The Italian singer Giusy Ferreri, best known for hits such as Roma – Bangkok, has applied to register a sound mark consisting of the spoken phrase “Sono Giusy Ferreri” – “I am Giusy Ferreri.” She joins Matthew McConaughey and Taylor Swift, who have filed similar applications in the United States, in a growing line of celebrities trying to use the trademark system as a shield against unauthorised digital replicas.
It is an eye-catching strategy. But will it actually work? And, more importantly for our clients at Trademarkroom, can ordinary business owners, performers, influencers and entrepreneurs do the same?
What is actually being registered?
A sound mark is precisely what it sounds like. Instead of registering a logo or a word, you register a noise. The MGM lion’s roar, the Intel jingle, and the Nokia ringtone – these are all classic examples. The Ferreri application takes the concept a step further by registering a spoken phrase performed in her own distinctive voice across classes 9 and 41, which cover things like downloadable recordings, audiobooks and entertainment services.
The theory is straightforward. If the phrase, in her voice, is on the register, then any AI-generated clip that mimics it could be challenged as trademark infringement. The application has passed the EUIPO’s internal review and will proceed to registration unless someone files an opposition by August 2026.
So far, so clever. The problem is that trademark law was not really designed to do this job, and the cracks start to show very quickly.
The distinctiveness problem
A trademark exists to tell consumers where goods or services come from. When you see the Nike swoosh on a trainer, you know who made it. The mark is acting as a badge of origin.
Now apply that test to “I am Giusy Ferreri,” spoken by Giusy Ferreri on an audiobook. Is the average consumer going to hear that phrase and think, “Ah, this product comes from the Ferreri commercial enterprise”? Or are they going to think, “This is an audiobook narrated by Giusy Ferreri”? Almost certainly the latter. The phrase describes the content of the goods, not their commercial origin. That is a classic distinctiveness objection, and it is the same hurdle that has tripped up many celebrity name applications before.
This issue matters for British business owners too. We see plenty of clients at Trademark Room who want to register their own name as a brand. It can be done – think Victoria Beckham, Jamie Oliver, and Stella McCartney – but it works because those names have become commercial identifiers through years of use on actual products. A name on its own, without that commercial backstory, is much harder to push through.
Substantial value and public policy
There is a second, more technical objection lurking in the EU Trademark Regulation. Article 4(1)(e) refuses registration to signs that consist exclusively of a shape or other characteristic that gives substantial value to the goods. The objection is intended to stop trademark law being used to monopolise things that should fall under copyright or design law instead.
Ferreri’s voice arguably provides substantial value to an audiobook in just that way. If you are buying “Giusy Ferreri reads her memoir,” the value of the product largely comes from her voice. The Court of Justice has made clear that substantial value is not just about aesthetics, and once a mark falls foul of this provision, there is no rescue – you cannot buy your way out of it through long use, as you can with distinctiveness.
Then there is the public policy ground. There is a reasonable argument that using trademark law to lock down a human identity stretches the system beyond what it was designed for. Trademarks are commercial tools; personalities are not.
Even if it registers, can you enforce it?
This is where the strategy really begins to falter. Let us assume, for the sake of argument, that the Ferreri sound mark is successfully registered. A deepfake video then appears on social media, in which an AI-generated Ferreri appears to endorse a dubious cryptocurrency. Can the trademark be used to take it down?
To succeed in a trademark infringement claim, you need a sign that is identical or similar to the registered mark, used in the course of trade, in a way that affects the mark’s functions. The registered mark is the phrase “Sono Giusy Ferreri” in her voice. If the deepfake says something completely different, even in a voice that sounds like hers, the registered sound mark is not really being used at all. You have a deepfake problem, but you do not necessarily have a trademark problem.
Even if the deepfake did reproduce that exact phrase, you would still need to show it was used in the course of trade. A meme account posting a joke clip is probably not trading. A satirical podcast may not be either. The infringement net, in other words, has some huge holes.
So what should UK business owners actually do?
Here is the honest answer we give clients at Lawdit. Trademarks are a useful weapon, but they are one weapon in a much bigger armoury. If you are a performer, an influencer, an author, a coach or any kind of personal brand, you should be thinking in layers.
The first layer is the conventional trademark. Register your business name, your stage name, your podcast title, and your signature catchphrase if it has been used commercially long enough to acquire distinctiveness. That is achievable, and it gives you a clear, enforceable monopoly across the goods and services you care about.
The second layer is copyright. Every photograph of you, every recording of your voice, and every video clip is potentially a copyright work. If a deepfake has been trained on, or directly incorporates, your protected material, that opens up an infringement route which is often much more straightforward than a trademark claim.
The third layer is passing off. The UK does not have a free-standing image right, but our passing-off doctrine has been stretched impressively far in recent years. Rihanna famously used it to stop Topshop selling T-shirts with her face on them, on the basis that customers were being misled into thinking she had endorsed the product. If a deepfake is being used commercially in a way that suggests false endorsement, passing off may be available even when trademark law is not.
The fourth layer is data protection. Your face and your voice are personal data under the GDPR. Unauthorised processing of that data, including training AI models on it, gives rise to rights and remedies that exist entirely outside the IP system.
The fifth layer is the developing patchwork of AI-specific regulation. The EU AI Act now requires deployers of systems that produce deepfake content to disclose that they have artificially generated the content. The UK government has indicated it is considering legislation to protect against unauthorised digital replicas and possibly a standalone image right. These are early days, but the direction of travel is clear.
The takeaway
Filing a sound mark for your own voice is a bold move, and it generates excellent headlines. It is not, on its own, a robust defence against deepfakes. Most of these applications will either be refused for clear reasons, fail due to opposition, or be registered but mostly unenforceable against the threats they were meant to defeat.
That does not mean trademarks are useless in this space – far from it. A properly drafted portfolio covering your trading names, your logos, your taglines and the commercial assets that genuinely function as badges of origin is still the foundation of any sensible brand protection strategy. But for everything else – the face, the voice, the persona, the likeness – you need to be thinking about copyright, passing off, data protection and the emerging AI rules in combination.
The Ferreri application is a fascinating moment in the evolution of intellectual property. It tells us, quite clearly, that the existing system stretches in ways its drafters never anticipated. It also tells us, just as clearly, that the answer is not going to come from any single area of law. It is going to come from a layered, joined-up strategy – the kind of approach that experienced IP solicitors have quietly been building for clients for years.
Need advice on protecting your brand?
If you are concerned about how deepfakes, AI-generated content or digital replicas might affect your business or personal brand, the team at Trademarkroom can help you put the right protections in place. We work with performers, influencers, authors, founders and SMEs to design trademark portfolios that actually work in the modern landscape – not just on paper.
Get in touch with us at tmr@trademarkroom.com or visit trademarkroom.com to arrange a no-obligation discussion with one of our specialist trademark solicitors.
