What does AI "disclosure" do to the value of music?

Over the past few months, Spotify, Apple Music, Tidal, Deezer and YouTube have each announced new AI-related initiatives. Spotify has launched AI disclosures within Song Credits. Apple Music has introduced Transparency Tags. Tidal has announced it will demonetize tracks identified as fully AI-generated. Deezer has built and licensed its own AI detection technology. YouTube has launched AI-powered tools that allow creators to replace music within their content to avoid demonetisation.

On the surface, these appear to be different solutions to different problems.

But in reality, they are all trying to solve the same challenge: How does the music industry preserve the value of human-created music in an ecosystem increasingly populated by AI-generated content?

The conversation is often framed around transparency. While transparency is important, it is only part of the issue. The larger questions are accountability and most importantly  economic. How is value attributed, how is compensation distributed, and who is responsible for ensuring that the system remains fair and workable in practice? 

Transparency cannot depend solely on self-reporting

Most of the current approaches rely heavily on uploader or distributor disclosures. Not only on DSP level, but also at source when registering the works with their publisher, PRO, and label. 

Many PROs, just like STIM and their registration process rely on the songwriter or publisher to disclose if and how much AI has been used. Spotify's AI disclosure beta allows artists to indicate where AI has been used within a recording. Apple Music's Transparency Tags similarly rely on labels and distributors to identify AI involvement. Tidal's policy also places significant responsibility on distributors when determining whether content should be identified as AI-generated. This also raises a broader question about incentives within the distribution ecosystem. Many distributors operate on business models that are driven by uploader volume rather than listening volume. DistroKid, for example, charges creators an annual subscription fee and allows them to retain 100% of their royalties. Its commercial success is therefore closely linked to attracting and retaining uploaders, regardless of whether those uploads are human-created, AI-generated, or somewhere in between.

As AI tools continue to lower the barriers to creating and distributing music, distributors are likely to see a significant increase in upload volume. While there is nothing inherently wrong with that, it does create an environment where the parties being asked to identify and disclose AI-generated content may not always have strong commercial incentives to actively police it.

This is why transparency cannot depend solely on self-reporting by songwriters, publishers, uploaders or distributors. The parties responsible for monetisation, recommendation systems and royalty allocation must also play an active role in verification and enforcement.

But before we can discuss disclosure, detection, compensation or demonetisation, we need to answer a more fundamental question: what exactly is AI-generated music?

Most AI discussions assume there is a clear distinction between human-created and AI-created works. In practice, the reality is far more nuanced.

If an artist uses AI to generate or edit already existing lyrics, is the resulting work AI-generated? What if AI is used to create parts of a melody that is subsequently rearranged, re-recorded and produced by humans?

What if AI is used to generate a vocal that is later mixed with human performances?

And what about the countless production tools that already use AI to assist with editing, mastering, mixing, noise reduction and audio enhancement?

At what point does AI stop being a tool and become a creator?

The industry currently lacks a shared definition, yet many of the proposed solutions depend on one. Without clear thresholds and consistent standards, disclosure requirements become subjective, detection becomes difficult, and compensation models become almost impossible to administer fairly.

DSPs must take ownership

One of the more striking aspects of recent announcements is how much responsibility continues to sit with uploaders. 

Yet DSPs are the parties that control discovery, recommendation systems, editorial playlists, monetisation frameworks and royalty allocation. They ultimately decide how content is surfaced, categorised and compensated.

If AI-generated content has the potential to influence those systems, then responsibility cannot rest solely with those uploading the content.

DSPs and PROs should be expected to establish clear detection standards, publish transparent methodologies, and provide meaningful appeals processes when content is classified as AI-generated.

Deezer has demonstrated that platform-level detection is technically possible. While no detection system will be perfect, the industry should move towards independently verifiable standards rather than relying primarily on self-reporting.

Transparency should not be something that platforms request from others. It should be something they actively help create.

What happens to the money? 

Perhaps the most important question remains largely unanswered. What happens when undisclosed AI-generated music enters the royalty ecosystem?

In a traditional pro-rata model, all streams contribute towards a shared royalty pool. If AI-generated content, not protected by copyright, captures a growing share of listening activity, that inevitably affects how revenue is distributed amongst human-created repertoire. It can also dilute the market share on which the pro-rata share royalty payout is based on. 

If a platform chooses to demonetize AI-generated content, what happens to those withheld royalties? Are they redistributed to human-created works? Are they retained by the platform?

Do they alter market share calculations?

Should previously distributed amounts be adjusted if content is later determined to be AI-generated?

These should not be theoretical questions. They are key to finding a solution that allows royalty flow.

For rightsholders, transparency around royalty treatment is likely to be more important than transparency around labelling.

A commercial opportunity? 

The discussion should not be framed solely as a defensive exercise.

The same infrastructure being built for AI disclosures can also create new commercial opportunities.

Verified provenance, auditable metadata and trusted disclosure standards could allow platforms, advertisers and consumers to identify and value human-created works differently. Human-authored catalogues, verified creator programmes and premium licensing products may all emerge from the same systems currently being developed for AI compliance.

The technology is here and continuing to develop, the business case becomes increasingly clear. So what remains is industry coordination.

The next phase

It's no longer about whether AI music should be labelled or if it's here to stay, but whether the industry can build systems that preserve transparency, maintain trust and continue to reward human creativity.

The music industry has spent decades developing frameworks to identify ownership, attribute usage and compensate rightsholders. AI challenges all three.

Transparency cannot rely solely on self-reporting and responsibility cannot sit entirely with uploaders,distributors and publishers/songwriters. Claiming, reporting and compensation models cannot remain an afterthought.

If DSPs control discovery, monetisation and royalty allocation, they must also take ownership of detection standards, transparency obligations and the economic consequences that follow.

Ultimately, the debate is no longer about whether AI is here to stay. It is. The challenge now is building an ecosystem that can distinguish between AI as a tool and AI as a creator, establish consistent standards for disclosure and registration, identify and filter AI-generated content at scale, and ensure that royalty distributions continue to reflect the value of human creativity. Until the industry can align on those fundamentals, questions around transparency, market share dilution, withheld royalties and fair compensation will remain unresolved. The opportunity is not simply to manage AI, but to build a more transparent and accountable music ecosystem than the one that existed before it.