AI-Generated Music Poses Legal Quandaries Around Transparency

A recent series of agreements involving artificial intelligence learning through a licensed “large music model,” as well as new EU regulations, provide guidance for how artists and music publishers can protect and enforce their rights in their creative works going forward.

Though AI-generated music may be able to mimic our favorite artists, debates persist over whether it can, or should, capture the soul or contain emotional depth of those artists. Regardless of where you land, the related legal issues are less abstract.

Many artists, music publishers, and other rights holders have claimed that AI companies are using their work to train AI models without authorization or compensation, leading to a wave of lawsuits that implicate core copyright doctrines. This raises legal and ethical concerns around transparency, human creativity, and artist autonomy.

Copyright Owner Rights

AI music creation implicates several exclusive rights granted to copyright owners.

Reproduction right: Training AI models often requires copying entire works or substantial portions into datasets, transforming them into machine-readable formats, and creating temporary or intermediate copies throughout the training process. Rights holders contend this constitutes copyright infringement unless a statutory defense, such as fair use, applies. Even temporary duplication of copyrighted text or audio can trigger infringement concerns.

Derivative works: AI-generated output may cross into unauthorized derivative work when it closely replicates the structure, cadence, melodies, or organization of copyrighted work such that it is substantially similar.

Distribution right: Platforms that permit users to generate and share AI-created audio files may implicate the exclusive right to distribute. Such a platform enables downstream public distribution and raises both direct and contributory liability concerns.

Figuring Out Balance

The recent Klay transactions and the European Union AI Act suggest a path toward balance.

Klay Vision Inc., a music technology company, recently announced that it has entered into separate AI licensing agreements with all three major music labels and their publishing arms.

Described as a first-of-its-kind arrangement, these deals allow Klay’s platform—powered by its large music model—to train exclusively on licensed music. Klay emphasizes that its platform doesn’t aim to replace artists with meme-based generation engines. Instead, it aims to provide an immersive, interactive experience where listeners can curate and enhance their musical experience while ensuring proper compensation for creativity of artists and songwriters.

The EU’s enactment of a comprehensive cross-sector AI statute helps establish binding transparency and compliance obligations that set minimum standards for creator autonomy.

For general-purpose AI, or GPAI, models, providers must comply with EU copyright law, including Article 4(3) of the DSM Directive, which permits copyright holders to expressly reserve their rights and opt out of text and data mining absent explicit authorization.

The EU Act also requires GPAI to “make publicly available a sufficiently detailed summary about the content used for training” following a standard template set by the EU AI Office. This framework allows for licensed, transparent models—similar to the Klay model—and provides guardrails for structuring future deals.

The Next Phase

While the Klay agreements and the EU AI Act mark significant progress, unresolved legal and ethical questions remain.

Who is the “author” of AI-assisted music, and which human contributions are copyrightable?

US law makes clear that copyright covers human authorship rather than machine-generated output. The issue is deciding when turning human input into generative AI is enough to count as authorship instead of technical support.

How should ownership, neighboring rights, and royalties be allocated among human creators, labels, publishers, platforms, and end users?

Existing allocation models assume clear roles between music creators and other rights holders, such as publishers and record labels. However, AI complicates these traditional distinctions by reshaping who can contribute creatively and how economic value is generated.

How will AI‑generated music compete with human-created music in the same commercial markets, and how can artists distinguish themselves?

For example, Breaking Rust’s AI-generated track “Walk My Walk” topped Billboard’s Country Digital Song Sales Chart, illustrating how AI-generated music can achieve mainstream traction. Similarly, Xania Monet, an AI-driven artist, has charted on Billboard and attracted a $3 million record deal.

Taken together, these developments underscore that AI-associated music is increasingly competing with human-created works for the same deals, chart recognition, and audience attention.

Should AI transactions warrant new royalty categories (for example, training, model-use, voice-model licensing), or fit into existing frameworks?

Existing royalty structures can be adopted to cover AI’s uses, which may preserve familiarity and predictability. But entirely new royalty categories may be justified, particularly for AI-specific needs, such as a data training.

What scope of consent applies? Can artists limit AI use to future works or specific projects, rather than their entire catalog?

Although AI developers may prefer blanket access to entire catalogs, artists and other rights holders will likely advocate for granular, ongoing consent that allows them to restrict, revoke, or tailor AI uses as technology and market practices evolve.

What to Know

Deals like Klay’s signal a shift toward licensed AI training through rights-cleared catalogs. Expect more opportunities, and more pressure, for AI companies to participate in structured licensing agreements rather than passive ingestion of source materials. While fair use remains in dispute, licensing agreements provide a more sustainable use case.

Artists and authors should review recording, publishing, and distribution agreements for clauses on new technologies, derivatives, text and data mining, machine learning training, voice likeness/right of publicity, and any clauses that let labels or publishers license on their behalf. Otherwise, though their rights may be clear today on one medium, such as streaming, their rights may be muddled, or non-existent on yet-to-be-developed mediums.

They also should formalize their ownership stakes contractually regarding scope, territory, revocability, labeling/disclosure, and prohibited contexts (for example, not using the tools for voice cloning or making political ads) to create enforceable parameters that protect both revenue and reputation.

To support enforcement, artists and authors should require audit-ready recordkeeping, similar to the EU AI Act obligations, to verify whether their work is being used in AI training and whether such use is within the bounds of the permission granted.

Outlook

While the introduction of AI into music creation has rightfully raised concerns around infringement, licensing arrangements such as the Klay deals and regulatory frameworks such as the EU AI Act demonstrate that innovation need not come at the expense of creators’ rights.

Indeed, the future of music not only involves AI but also includes licensed training, usage transparency, and robust consent mechanisms that preserve creative and economic agency for artists.

This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law and Bloomberg Tax, or its owners.

Author Information

Daniel A. Rozansky is a partner at Brown Rudnick with decades of experience as a commercial litigator, with significant emphasis on brand and reputation management in the media and entertainment industry.

Berlindyne B. Elie is an associate at Brown Rudnick in the litigation and dispute resolutions practice.

Write for Us: Author Guidelines