AI deepfakes and the music industry

The music industry is no stranger to third party challenges. From teenagers creating cassette mix tapes in their bedroom to the evolution of online sharing in the form of a digital revolution triggered by Napster and ongoing evolution in this area through legitimate and illegitimate means, through music listening services and platforms such as YouTube, the music industry has had to maintain a creative edge in order to ensure that artists’ rights can be adequately protected.

The world is now faced with the rapidly evolving challenge of artificial intelligence. This is a particular issue for the music industry in terms of the ease with which new material can be created without any involvement from the original artist at all. Whilst many welcome the more positive benefits of AI as a production tool and reducing costs in terms of numerous aspects of creation and development, it brings with it a significant issue in terms of potential copyright infringement including in the previously niche area of voice cloning. The speed and sophistication of the AI tools now available mean that it is relatively easy to generate voice files, including songs, that have nothing to do with the artist they are supposed to sound like.

Why artists are protecting their voices

Any advance in technology requires creativity to find ways to tackle the negative aspects on a cost-effective basis. This has led to a recent trend of artists seeking to protect their voices.

Celebrity voice marks and sound mark filings

Lionel Richie has filed four US trade marks to protect the sound of his voice speaking four of his iconic phrases: “Hello, is it me you’re looking for?”; “Easy like Sunday morning”; “All night long”; and “Say you, say me”. The intention behind these filings is not necessarily to protect the song lyrics, but rather to protect Richie’s voice.

These “sensory marks” are filed in class 41 for various entertainment services on an intent to use basis, meaning Richie will need to prove use of these trade marks in commerce before registration is granted. Will consumers view these marks as indicators of origin or merely as lyrics forming part of well-known songs? The use obligation is not present in the UK or EU (although the UK requires a bona fide intention to use at the date of filing), but the marks must still be perceived as indicators of commercial origin.

Richie’s trade mark filings follow those of Taylor Swift who looked to protect her voice saying “Hey, it’s Taylor” and “Hey, it’s Taylor Swift”, which are also pending in the US. Swift has previously been the victim of deepfakes and, being IP aware, has a range of trade mark applications and registrations to protect her rights.

Matthew McConaughey has been equally creative filing various US applications for audio and visual clips to seek to cover his voice and image.

In Europe, Italian singer Giusy Ferreri, filed a European Union trade mark (EUTM) application for a sound mark consisting of the phrase “SONO GIUSY FERRERI” which has been published, meaning the EUIPO has accepted that it has met the requirements for registration in terms of distinctiveness. Ferreri's filing is particularly notable because it explicitly targets protection against unauthorised digital replicas and deepfakes. The application demonstrates an emerging strategy: filing voice marks not as abstract vocal characteristics, but as the artist's voice reciting a distinctive phrase closely associated with their identity.

Image, motion and personality trade marks

In the UK, those in the public eye have also been looking at how to protect themselves in a rapidly evolving landscape. The darts champion Luke Littler has registered a passport-style photograph of his face for a broad range of goods and services, including those seemingly unrelated to his darts career. In Europe, the EUIPO Board of Appeal (Johannes Hendricus Maria Smit v EUIPO, case R 50/2024-2) has yet to decide whether personality-based marks including a photorealistic image of a human face can function as a trade mark. The anticipated judgment will likely shape the future framework for protecting personal identity as a trade mark, potentially influencing voice mark strategy as well.

Chelsea FC footballer Cole Palmer registered a UK motion mark in 16 classes described as “…a short video sequence, approximately 1.5 seconds long, which depicts the professional footballer, Cole Palmer, with his arms crossed across his torso, performing a shivering motion by rubbing his hands up and down his upper arms repeatedly”.

It is unsurprising that music artists, actors and other celebrities are looking to protect their valuable appearance and voices in the age of AI.

AI-generated music and deepfake takedowns

Deepfakes and fake music tracks have caused issues for numerous individuals. In 2023, a track was released that featured convincing AI vocals mimicking the voices of Drake and The Weeknd, becoming a viral track that generated millions of streams across platforms before being taken down.

In March 2026, it was widely reported that Sony Music asked streaming platforms to take down more than 135,000 songs created with AI that impersonated some of its biggest acts. At the time Sony expressed concern that such deepfakes would damage release campaigns or tarnish the reputation of the artists. For listeners, they are faced with a world where it can be difficult to be certain that their preferred artists created what they are listening to.

How platforms are responding to AI music

In response, some platforms are looking to implement a process for approval of new releases to prevent future issues. This will not stop AI being misused across all platforms, but proactive steps such as this combined with a sensible IP protection strategy provide tools for taking action.

Platforms and music labels are also becoming creative in their embracing of AI, allowing it to be deployed in certain circumstances to satisfy the public desire to utilise AI and produce bespoke creations. In May 2026, Spotify partnered with Universal Music Group to let premium Spotify subscribers legally generate AI remixes. This offered a benefit to the participating artists in the form of new revenue streams for musicians as well as, of course, record companies.

In a constantly-changing landscape, trade marks are more important than ever, whether to enforce against copyists (AI-generated or otherwise), or to provide additional licensing revenue for artists and record companies.

Choosing goods and services for sensory marks

Before filing any trade mark it is key to consider the goods and services it will (or may be) used on. This is more difficult with sound or motion marks, where they cannot be physically be applied to goods. How do they appear at the point of sale of a product or offering of a service? Richie’s US sound marks are fairly narrow in scope, covering class 41 entertainment services only.

As mentioned above, in the UK and EU, there is no requirement to demonstrate use to achieve registration, which may provide greater flexibility for the scope of goods and services to be covered, but it is important to consider what goods and/or services are realistically likely to be offered under the mark and, with an element of crystal ball gazing, what consumers may like to see on the market.

Entertainment services, sound and video recordings, merchandise, digital and virtual services…the list is long and will need to be considered on a case-by-case basis.

How UK and EU IP offices and courts may treat sensory marks

As always with unusual trade marks, there are questions over whether they are inherently registrable and also how they will (or can be) enforced once registered. How well understood will the scope of the rights afforded by such marks be by third parties, the registries and/or the courts? How easy will it be to decipher whether a particular instance of unauthorised use is an infringement or just happens to sound similar? As ever, time will tell, but this will be a fascinating landscape for the next few years. Forward-thinking innovative artists and record companies are at the vanguard of pushing the boundaries of brand protection within this brave new world, and will undoubtedly lead to positive developments within and growth for the music industry, which remains an important creative core of our society.

The first hurdle is persuading relevant registries that the mark is inherently registrable, that it is distinctive enough to identify the origin of the goods and services covered, and not merely descriptive. How will that mark function as a trade mark in the course of trade and how will it be viewed by relevant consumers?

Enforcing voice marks against AI misuse

Once registered, how will it be enforced? It may be a challenge to assess similarity of one sound mark to another, especially if the words spoken or sung are not identical. How will similarity of other personal attributes be assessed? Will the infringement be viewed as merely descriptive use or otherwise not in the course of trade?

The marks will need to be assessed within the existing legal framework, which was not drafted with AI and other more recent issues in mind. Perhaps a close case is that of the then European Court of Justice (ECJ), now the Court of Justice of the European Union (CJEU), in L'Oréal SA and Others v Bellue NV and Others (case C-487/07), which considered smell-alike perfumes and the use of comparison tables to the original fragrance, which created a link between the original and the imitation in the minds of the public and took unfair advantage. There is a logical extrapolation to be made here, and sound-alike cases may well be on the horizon in the near future.

Arsenal Football Club plc v Reed ([2003] EWCA Civ 696) confirmed that even goods stated to be unofficial can be an infringement and that claiming a mark is merely an indicator of fan loyalty or a badge of allegiance does not escape liability.

Building a holistic AI brand protection strategy

In a world where artists and other rights holders in the creative industry have to consider how best to defend themselves against malignant use of new technology, a bespoke, holistic strategy is required. Novel applications for the sound of voices, key lyrics, song names, album titles and motion marks to cover “signature moves” of celebrities should be combined with traditional word and logo trade marks for broad protection. The potential to register song or album names, lyrics, fonts and album cover artwork, as trade marks, to sit alongside any copyright or design protection in place, should all be considered. A sound trade mark protects against commercial exploitation and misuse as an identifier, particularly where deepfakes or voice cloning occur in commerce.

Such applications may encounter issues with distinctiveness, and the honing of a proactive, strong portfolio requires creativity and careful thought. Overreaching specifications risk refusal; a too narrow scope leaves gaps.

Trade mark strategy for emerging artists

As new artists enter this increasingly dynamic, AI-driven world, they should look to supplement their brand arsenal with bold trade mark filings that offer them a competitive edge, and may even give them a degree of bargaining strength when negotiating record deals or other commercial agreements. If the right protection is put in place at an early stage of artists’ careers, they can legitimately gate-keep the use of their name, and potentially even their image and voice, with a view to generating additional revenue streams from licences which deploy those distinctive identifiers with a guarantee of authenticity.

Creative brand protection in the AI era

Budgets will inevitably be tight, but with the right guidance from trusted advisers, strong brand foundations can be put in place at a reasonable cost. Being creative, appropriately, is key. As is often the case, inspiration can be sought from the great Lionel Richie, clearly as vexed as anybody else at having to search for his bags on a busy airport carousel. He has registered the mark HELLO, IS IT ME YOU’RE LOOKING FOR? in the US in relation to “luggage tags”.