AI Music: Rights & Copyright
Artists' protest against AI-generated songs is gaining momentum. Paul McCartney contributes a silent track to a protest LP against the unregulated use of AI. Earlier, over 1,000 artists had released a "silent" album, with proceeds going to Help Musicians. The central question is how creativity, technology, and fair regulations can be reconciled without musicians having to bear the costs.
Source: Own Representation
Artist Protest & AI Music
AI-generated songs are created by generative models that learn patterns from large datasets and derive new sound sequences, styles, or voices from them. Training these models on copyrighted works without consent and compensation is legally controversial and is being debated in the US and Europe. The EU requires providers of general-purpose AI models to provide a public summary of their training content, for which the Commission published a template in July 2025.
On February 25, 2025, "Is This What We Want?" was released, an album featuring recordings of empty studios, supported by over 1,000 creatives such as Kate Bush, Damon Albarn, and Hans Zimmer., The track titles of the album form a political message to the British government. The project page lists contributors, fundraising goal, and information about the vinyl release. On November 17, 2025, it was announced, that Paul McCartney is providing an additional "silent" contribution for the vinyl edition. Labels like Universal Music Group are trying to establish new business models. At the end of October 2025, UMG settled a legal dispute with AI provider Udio and announced a licensing partnership. . Other proceedings, such as against Suno, are ongoing.
The artist protest has several motives. First, it's about consent: many demand an opt-in for training data instead of an opt-out. Second, it's about compensation: if AI outputs resemble originals, revenues and commissions can be displaced. Third, transparency is demanded. The EU is trying to counter this with a mandatory training data summary, but is criticized because details and enforceability remain unclear. Platform dynamics amplify the pressure, as the Drake/Weeknd deepfake of 2023 showed, which quickly went viral before rights were clarified.
Source: tonstudio-wissen.de
The question of 'curse or blessing' accompanies the discussion about artificial intelligence in music.
Source: YouTube
Legal Aspects & Transparency
The silent album exists, and proceeds go to Help Musicians. A vinyl edition with a bonus track by Paul McCartney is planned. On July 24, 2025, the EU published an official template for training data summaries. UMG und Udio haben ihren Streit beigelegt und eine licensing partnership announced.
It is unclear whether training on protected works without a license is considered "fair use" in the US. The proceedings against Suno and Udio, among others, have not been decided by the highest courts. It is also unclear how detailed EU models must disclose their data usage and what this means for individual rights holders.
The claim "AI music is illegal per se" is false. AI tools are allowed; what matters are usage rights, the origin of training data, and human contribution. Even the Grammys allow works with "substantial human input." The statement "platforms are doing nothing" is also misleading. YouTube has allowed takedown requests against unauthorized face/voice simulations since 2024 and requires disclosure for realistically altered content.
Source: beat.de
The rise of AI in the music industry is leading to tensions and protests among artists.
Artist associations and musicians criticize opt-out models and demand licensing with compensation. The EU refers to the Code of Practice and the new template as steps towards greater transparency. AI companies argue for innovation and fair use; some are now moving towards licensing models, partly after settlements.
Impact for Musicians
For bands, producers, and composers, this means documenting their rights, setting up registrations and contracts cleanly. On YouTube, claims should be substantiated and disputed claims handled in a structured manner. Misuse should be reported; on YouTube, deepfake voices can be removed via a privacy process, and synthetic content must be declared.
Discoverability can be increased by fingerprinting services like Pex or Audible Magic that track usage. Udio intends to fingerprint generated tracks as they are created. The origin of content should be labeled. Content Credentials/C2PA make editing steps traceable and are increasingly supported in audio workflows.
Source: blog.suisa.ch
The connection between artificial intelligence and copyright raises complex legal questions.
An example of declaring synthetic content on YouTube could look like this:
<meta name="youtube:synthetic-content" content="true">Source: YouTube
Open Questions & Conclusion
It remains open how detailed the EU templates will be filled out and whether they will provide reliable information on specific datasets from which rights holders can derive claims. In the UK, it is unclear whether opt-in or opt-out will ultimately apply and what fair compensation could look like in practice. In the US, the fair use question for AI training remains unresolved; the Copyright Office's analysis shows areas of tension but no final line.
The artist protest against AI-generated songs is more than symbolism. It forces politicians, platforms, and companies to specify rules for consent, transparency, and compensation. For musicians, this means securing rights, gathering evidence, using reporting mechanisms, and simultaneously exploring opportunities where licensed AI tools can enhance creative processes.