How UK’s AI Copyright Changes Could Impact the Music Industry

More than 1,000 musicians have released a silent album protesting the UK government’s proposed changes to the copyright laws, allowing companies to train their AI models on copyrighted works like music, art, and text without a license or remuneration.

The album titled ‘Is This What We Want?‘ is live on Spotify and comprises empty recordings of studios and performance spaces, which the group claims represent the impact of the government’s proposal on the musicians’ livelihoods. Interestingly, the titles of the different tracks on Spotify spell out the message “The British Government Must Not Legalise Music Theft To Benefit AI Companies”.

In a separate letter to The Times, several artists, including Elton John, Dua Lipa, and Paul McCartney, warned that the proposal undermines UK copyright law—a key factor in attracting rights holders and investment. Substantiating their claims, they added that the UK’s creative industries contribute 126 billion pounds to the country’s economy annually and employ 2.4 million people.

The UK Government’s Proposal on Copyright Laws

Announced for public consultation in December 2024, the UK government’s proposal aims to curb the worries of rights holders expressing concerns about their works used in training AI models and those of AI developers facing issues in navigating the UK’s copyright law. The plan is based on three objectives- supporting rights holders’ control over their content, aiding the development of “world-leading AI models by the UK”, and fostering greater transparency.

While opening the proposal for public consultation, the government outlined three different options, while mentioning its preference for each mode:

STRENGTHENING EXISTING COPYRIGHT PROVISIONS

This method requires licenses for AI companies to train their models on copyrighted works, backed by transparency provisions and easier routes to enforce copyright. However, the government contends that while this would improve access to remuneration for creators, it would create difficulties for AI companies while accessing content in the UK when compared to other jurisdictions. Finally, since the option curbed access to material for AI developers, despite giving right holders greater control over their works, it wasn’t given preference.

broad data mining exception- a free license?

Within this option, companies could openly mine data of copyrighted works for AI training purposes without the right holders’ permission. This mechanism would be subject to certain exceptions like those listed in other countries like Singapore and the U.S. However, since the approach fell short of transparency and control objectives, it was not preferred.

data mining with rights reservation

This preferred suggestion would enable companies to undertake text and data mining of copyrighted works unless the right holders proactively opt out. Additionally, online works should include a machine-readable reservation of rights, which would automatically indicate copyright restrictions to AI systems. The government claims that this method satisfies all the aforementioned objectives and provides for “spillover innovation and productivity benefits”, enabling the UK’s AI sector to become more competitive internationally.

This proposal signalled part of the country’s broader AI strategy, termed the “AI Opportunities Action Plan”, which detailed investments like dedicated “AI Growth Zones” for infrastructure development and funds worth 14 billion euros for this purpose. This sharply contrasted with the former British government’s cautious approach toward AI, which focused on tackling its risks.

Why are artists worried?

While tech giants like Google have previously argued in favour of the “opt-out” model for training their algorithms, publishers have expressed concerns over increased compliance burden, the Financial Times reported. Fearing “widespread theft of copyrighted material without remuneration”, creators have also pointed out flaws like unawareness about which companies are trying to mine their content within this system. Generative AI companies circumventing their responsibilities was another issue highlighted within this proposal.

Advertisements

Besides musicians, major newspapers across the UK voiced opinions against the proposed efforts displaying a prominent “MAKE IT FAIR” message on their front pages. The campaign appeared in publications on the same date as the end of the AI and copyright consultation, i.e. on February 25, 2025. Although the two instances differ in context, this campaign by UK media outlets is reminiscent of the Indian Express’ famous blank editorial on June 28, 1975, which protested press censorship during India’s Emergency. To explain, following the resuming of its publication, the newspaper ran a blank editorial page highlighting press censorship by the Indira Gandhi government, drawing attention to injustice, akin to what UK publishers seek to do.

Relationship between music labels and AI companies

Previously, music labels like UMG, Warner Brothers Music and Sony Music have filed complaints against AI companies for training their AI models from sound recordings from the labels’ copyrighted works. Conversely, other companies like YouTube have also initiated discussions with record labels to “legally obtain popular artists’ songs to train its new AI tools”.

Meanwhile, artists have called upon stakeholders to block AI models from ‘infringing upon and devaluing the rights of human artists’. Previously, the American Union of Media Professionals- the Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA) in a tentative agreement with record labels has also demanded guardrails for artificial intelligence. This comprised the establishment of clear and conspicuous consent, minimum compensation requirements, and specific details of intended use before releasing a sound recording using a digital replication of an artist’s voice.

The copyright conundrum

Generally speaking, music copyrights are of two kinds– songwriting or publishing copyright (the rights over the songwriting or composition of a music piece) and sound recording or master copyright (rights belonging to the owner of the master sound recording). Speaking of deals with record labels, they own and control all elements of the copyright. Further, the copyright system is key to receiving royalties or payments made to owners when the music is played, performed, or reproduced.

Considering the UK government’s new proposal, since the copyright holders (record labels in certain cases) can prevent AI training based on their work via an “opt-out”, it still leaves artists and songwriters out of the picture. Consequently, their rights over content created by them may end up being trampled.

Music labels on USing AI

This concern may also be amplified by the view of music labels towards AI. For instance, UMG’s Senior Vice President for Strategic Technology Chris Horton remarked about the future alignment of licensing, litigation, and legislation to enable AI companies to work with creative industries in an interview with Music Ally. He also added that “music will be more interactive and more responsive” and that “musicians will find incredible uses for AI systems we can’t predict today”. In contrast, Sony Music Group considered embracing the use of AI as a creative tool, it opted out of AI training, given the wishes of artists and songwriters. Overall, with the disparate views of these record labels and the general criticism of the UK’s proposed copyright changes, it remains to see how artists adapt to the same and the subsequent steps they take.

Also Read: