“Is it Jorja Smith on that track?”: AI and the Music Industry – a growing copyright challenge

Identifying whether a track was created by artificial intelligence (“AI”) is becoming increasingly difficult. This is a concern not for listeners but for artists, rightsholders, collecting societies and other stakeholders. As AI expands its influence on the music industry, issues of ownership, originality and infringement in copyright are becoming urgent. There are also issues that could arise under the laws of passing off.

This summer, the AI-generated band Velvet Sundown climbed into the UK’s viral Spotify charts, signalling a significant shift. Recently, award-winning artist Jorja Smith’s label, Famm, publicly criticised British dance act Haven and the rise of AI-produced music generally. This stemmed from the accusation that the AI generated track produced by Haven was trained on Smith’s music. Famm stated:

“This isn’t about Jorja… It’s bigger than one artist or one song.”

The reality is that AI’s entry into the music scene looks set to trigger a wave of intellectual property disputes, with copyright, the cornerstone of the industry, at the centre of the battle.

This blog examines the growing influence of AI in the music industry. As AI-generated tracks climb the charts, the legal framework safeguarding music is under increasing scrutiny. The recent Getty Images (US) Inc & Ors v Stability AI [2025] EWHC 2863 (Ch) case offers some insight, although that decision left many wanting greater clarity, as key copyright issues were dropped over its course. Just as image generators trained on copyrighted works raised questions of liability, AI music generators trained on protected tracks will face similar challenges. If outputs substantially reproduce original compositions, who bears responsibility—the developer or the user?

How is Music Currently Protected?

Under UK law, musical works benefit from automatic copyright protection as soon as they are created and recorded. There’s no need for registration – with originality and fixation being key. Copyright safeguards several distinct elements including:

1. Musical composition – melody, harmony, and arrangement
2. Lyrics – treated as literary works
3. Sound recordings – separate rights that apply to the recording
4. Printed editions – the layout of sheet music

Artists also develop a reputation, and goodwill over their careers in the music industry. This can be recognised under the laws of passing off, as seen in the case of Robyn Rihanna Fenty v Arcadia Group Brands Ltd (T/A Topshop) [2013] EWHC 2310 (Ch). This allows artists a certain level of protection against unauthorised parties deploying their name or image to offer goods or services to the public.

Licensing and Enforcement

Usually, in order to use any of these works – whether this is remixing, performing publicly, broadcasting, or streaming publicly – a licence is required in order to avoid infringing works. Licensing also ensures royalties flow back to creators and many artists join collective management organisations such as PRS for Music and PPL, which administer licences and protect rights on behalf of members.

Jorja Smith’s Record Label Challenges AI-Generated Music

As mentioned, Famm, raised serious concerns about the growing influence of AI in music production. The dispute centres on the track I RUN (“the Track”), created by producer Haven, which has gone viral on TikTok and amassed millions of views on Spotify and YouTube. Famm alleges that the Track was generated using AI trained on Smith’s work and is now seeking compensation. The Track was posted on multiple platforms and social media posts were tagged with “#jorjasmith”, which Famm claim were attempts to “circumvent usual takedown procedures”, and cause “public confusion as a key part of the marketing strategy”.

According to Famm, this case represents a pivotal moment for the music industry’s relationship with AI. Haven claims the vocals are their own but admits they were heavily processed using Suno, an AI music platform. Suno has acknowledged that its software is trained on copyrighted material, arguing that this falls under “fair use”. In the UK this refers to provisions in the Copyright Designs and Patents Act 1988 (“CDPA”) permitting limited use of copyrighted works for purposes such as criticism, news reporting, research and private study, and incidental inclusion.

The Track was set to chart but was removed from charts and streaming platforms after takedown notices were issued by Famm and industry bodies including the Recording Industry Association of America and the International Federation of the Phonographic Industry. Famm claims that, after the Track’s success, the producers invited Smith to feature on a remix, an attempt, they argue, to legitimise a track that had misled listeners into believing it was her work.

Although the producers have since altered the Track and removed hashtags referencing Jorja Smith, it remains unclear whether her recordings were part of the AI training data. Haven maintains they simply prompted the software to use “soulful vocal samples”. Famm “still believe both versions of the track infringe on Jorja’s rights and unfairly take advantage of the work of all the songwriters with whom she collaborates.”

Famm states it is pursuing royalties but has not specified whether this is via legal proceedings or commercial negotiations. The references to reputation and misleading in Famm’s complaints indicate that there could be a claim, not only in copyright infringement, but also in passing off. The hashtags used, and alleged deliberate prolonging of the public misconception that the Track did feature Jorja Smith, could be found to be attempts to mislead the relevant public that the Track was somehow connected to Smith.

From interviews with Famm and the facts revealed about the case, the following issues could arise:

• Have Haven infringed Famm, Jorja Smith, and other songwriters’ copyright in using Suno to create the Track?
• Have Suno infringed the same copyright in training their system?
• Have Haven passed themselves off as Jorja Smith by (a) creating the Track, and/or (b) using her name in its promotion?
• Is there a possibility that Suno could be passing themselves off as Jorja Smith, by way of providing the means for the Track?

Such a claim is likely to involve questions previously unanswered in the courts (even following the Getty judgment) or under statutory law:

• Does training a generative-AI system on copyright works without the author or rightsholders’ licence constitute copyright infringement?
• Is it the user or the platform that is responsible for acts of IP infringement that result from general use of the platform?
• Do any existing permitted acts in the CDPA apply to training gen-AI platforms, and if not, will an equivalent provision be applied for training such models, particularly when commercialised?

Famm has called for clear labelling of AI-generated music, enabling consumers to make informed choices. They also advocate for proper credit and compensation for artists whose works are used to train AI systems.

In the Getty case many of the copyright issues in the original claim fell away over the course of proceedings, due to jurisdictional issues related to the fact that the act of copying took place outside England & Wales where the servers were based.

However, it may have shed some light on who would be found responsible for infringing outputs in such a context. Stability, the platform developer, sought to argue that individual users were responsible for the output of the Getty mark on images generated by the platform (in a trade mark use context). The Court however disagreed, finding that it is Stability’s training of the model that causes the mark to appear.

The facts and claims in that case were somewhat different, Getty brought proceedings against Stability because its systems were throwing up images with the Getty marks on it, which are registered trade marks. However, in the context of generative-AI in music, this begs the question of whether claimants will be able to pursue the platform developers for infringing outputs.

In Smith’s case, should Famm and other claimants be bringing proceedings against Suno primarily? Is Suno responsible for generating the output that became the Track, more so than, or at least equally to, Haven? These questions are likely to arise if Famm initiates proceedings, and even if they do not, similar issues are expected to emerge in other cases.

Suno’s Previous Legal Battles and Industry Shift

Last year, Suno faced a major lawsuit in the United States from Sony Music, Warner Music Group (“WMG”), and Universal Music Group for copyright infringement. The labels alleged that Suno scraped vast amounts of music from the internet, including their catalogues, to train its AI model. They argued this amounted to direct infringement because the model was learning from copyrighted works. The damages sought were significant, up to $150,000 per infringed work.

However, the landscape shifted last month when WMG signed a licensing agreement with Suno. As part of the deal, WMG dropped its lawsuit and agreed to collaborate with Suno. Under this arrangement, WMG artists can now opt in to allow Suno to use their music for AI training. In return, WMG gains control over the output and receives licensing fees. This marks a critical moment, as rights holders negotiate commercial deals for consent and compensation rather than relying on the courts to provide legal clarity on the issues.

AI scraping is notoriously difficult to detect, making enforcement challenging. WMG’s approach could become a model for the industry, balancing inevitable innovation with artists’ rights, outside of the legislative sphere. Other major labels may follow suit, but in each case, it depends on whether the parties can reach an agreement that suits them both.

Legislative Developments

The government has been working towards establishing a legislative framework for AI and copyright. We previously explored the UK Government’s consultation on this here. However, no final response or decision has yet been issued.

The proposals included a text and data mining (“TDM”) exception, whereby AI developers would be exempt from copyright infringement (as long as access to copyright works was lawful). Connected to this was the ability for rightsholders to “opt-out” of their works being used for such training. However, this proposal was largely received as giving developers carte-blanche access to works, whilst placing the responsibility for opting out on the rightsholder.

The music industry has been vociferous in their opposition to the proposals, with the latest development being Paul McCartney releasing a silent track in protest. Recently there has been a shift in government tone about the approach, and whilst a TDM is still being considered, it may not be as wide-ranging as first discussed.

An ongoing inquiry, launched in November by the Communications and Digital Committee, is examining how creative rightsholders can meaningfully reserve and enforce their rights in relation to AI systems.

Evidence sessions have been taken place with another scheduled for this week. Previous sessions including one on 4 November 2025 featured Tom Kiehl, Chief Executive of UK Music, whose members include major producers. Kiehl highlighted the economic and cultural significance of the music industry, stating:

“The music industry is worth £7.6 billion to the economy. It generates £4.6 billion in exports. It employs over 216,000 people. Beyond the commercial impact, it has profound cultural and soft power influence. The copyright framework is fundamental to the music industry, both from a business perspective in terms of investment and for creators who derive value from their intellectual property. That framework underpins the success that powers the £7.6 billion.”

Concluding thoughts

This issue will continue to evolve, and both government and courts will need to establish a clearer path forward. AI has unprecedented capabilities for composing and producing music, generating lyrics, and mixing and mastering recordings with little to no human input.

Listeners and users do not necessarily identify that a particular track may be produced entirely using a gen-AI tool, and in the meantime, those tools are being trained on existing artists’ past work. The fact that some AI-generated tracks are reaching the charts puts the issue in the forefront, and such cases are surely set to increase.

Ultimately, there is a pressing need for some mechanism to protect artists’ copyright and ensure artists and rightsholders are properly compensated for creative content in the age of generative AI. The role of government and the courts will be to balance innovation with the rights of creators.

We will continue to monitor developments in the use of generative AI within the music industry and assess how legal protections evolve. It remains to be seen whether this will be driven by legislative intervention, the courts or the industry itself, as demonstrated by recent actions taken by Warner Music Group.

If you are a rightsholder that has discovered tracks you believe infringe your rights, get in touch with Lakmal Walawage, or Ruby Clarke, to discuss how you may be able to safeguard your work.