Àrèmúorin™ | Award-Winning Jazz, Soul & Inspirational Artist

YouTube Music has begun rolling out labels for tracks created primarily using generative AI. Transparency is welcome. But a label answers only one question: Was AI involved?

It does not answer the questions that matter just as much: Whose work helped make it possible? Was permission obtained? Was the songwriter identified? Was the recording licensed? And who gets paid?

That distinction may define the next stage of the AI music debate.

Disclosure Is Not Compensation

There has been enormous discussion about identifying AI-generated content. Identification matters. Listeners deserve to understand what they are hearing, and creators deserve protection against confusion between human work, AI-assisted work and entirely generated material.

But we should be careful not to mistake disclosure for accountability.

Putting an AI label beside a piece of music does not resolve the economics underneath it. If copyrighted songs or recordings contributed to an AI system, the deeper question remains whether the people who created those works participated in the value subsequently created.

A transparent exploitation can still be exploitation.

The Industry Is Moving From Litigation to Licensing

Something significant is already happening. In August, BMG and Suno announced a global alliance covering BMG’s recorded and music publishing repertoire. The companies said participating artists and songwriters would have their rights protected and be compensated, while the agreement also settles prior use of BMG recordings and publishing works.

These developments suggest that the argument is no longer simply: AI or no AI?

The more useful question is becoming: What does a fair AI music economy actually look like?

That is progress. But independent creators must be inside that conversation too.

The Song Cannot Become Invisible

There is another danger beneath all the excitement surrounding models, voices, recordings and generated outputs. The composition itself can disappear from view.

A recording has value. A voice has value. A performance has value. But before many recordings existed, somebody wrote a melody, lyric, harmony or structure that made the music recognisable in the first place.

On 30 September, independent publishing organisations IMPF and IMPEL released principles for generative-AI licensing. Their central position is particularly important: absent other significant factors, the song should be valued at least equally with the recording. Their framework also calls for transparency throughout the licensing process and meaningful rightsholder oversight.

That is not resistance to technology. It is an attempt to build functioning economics around it.

AI-Assisted Is Not the Same as AI-Generated

We also need greater precision in the language. There is a profound difference between an artist using technology somewhere within a creative process and pressing a button that generates an entire work.

Musicians have always used technology. Studios changed music. Synthesisers changed music. Sampling changed music. Digital recording changed music. Software changed music. AI will change it again.

The important questions are therefore not merely whether technology was present, but where human authorship occurred, what the technology contributed and whose pre-existing intellectual property was used.

We need vocabulary sophisticated enough to distinguish assistance from replacement.

Independent Artists Need More Than Principles

Major rights holders can negotiate directly with technology companies. An independent songwriter with a catalogue scattered across distributors, collecting societies, databases and historic registrations may have considerably less negotiating power.

That is why the infrastructure matters. If AI becomes another enormous exploitation layer on top of streaming, creators should not have to spend years discovering where their work went, which database failed to match it or which intermediary collected money they never saw.

Technology capable of analysing millions of recordings should also be capable of maintaining meaningful provenance. AI should make attribution better, not more mysterious.

From “Made With AI” to “Paid Through AI”

Perhaps the most useful evolution of the new label would therefore be conceptual. Today we are beginning to ask platforms to tell us: Made with AI?

Tomorrow creators should be able to ask: Licensed from whom? Attributed to whom? And paid to whom?

Imagine an AI music ecosystem where provenance travels with the work; authorised repertoire can be identified; songwriter, performer and recording rights remain distinguishable; and compensation can follow actual usage.

That would be considerably more ambitious than simply attaching a badge to a song. And considerably more valuable.

Innovation Does Not Require Amnesia

I believe emerging technology can expand creativity. It can help independent artists experiment, produce, arrange and imagine things that once required resources beyond their reach. That possibility deserves protecting too.

But innovation should not require the creative industries to forget where music comes from.

Behind the dataset is a catalogue. Behind the catalogue are songs. Behind the songs are human beings.

And if those human beings contributed value to the technology of tomorrow, transparency should tell us more than the fact that AI was involved.

It should help ensure that the creator remains visible — and that the creator participates in the value created.

Aremuorin
Thoughts. Culture. Truth.