You are not a creator simply because you have the tools. And having the tools certainly does not mean that everything generated through them belongs on streaming platforms.
After attending the closed session with Berlin AI Think Tank and Musikförläggarnas Industry Day, we came away with a rather positive outlook on the future of music and AI. Some might even call it optimistic.
But before we explain why, we need to address a fundamental question: can creative prompting constitute human authorship, and can the resulting AI-generated work qualify for copyright protection?
Is Prompting Creative?
In February 2026, the Munich Local Court considered this question in AG München, Endurteil v. 13.02.2026 – 142 C 9786/25.
The case concerned AI-generated logos, where the claimant argued that his prompting process constituted sufficient human creativity to establish copyright protection.
The court dismissed the claim, finding that the human contribution did not meet the threshold for copyright protection.
However, the court did not rule out the possibility that prompting could contribute to copyrightable output. Instead, it identified three essential considerations:
Human creative influence is essential.
Iterative prompting can be relevant.
The creativity must be reflected in the output.
The critical question is whether the human has exercised sufficient free and creative choices to make the resulting work their own original creation, this is where things become very interesting.
When we use generative AI, how much creative control do we actually have? We may think we have control due to specific prompting, but inn many cases, we are not making the creative choices ourselves. We are presented with choices generated by the AI. The model draws on patterns and relationships learned during training, generates an output and allows us to accept, reject or refine the result, through iteration. During our panel at IMPF in Palma, Sureel AI shared that, when they followed the process of how the AI generated an output, the prompter would give 2 writers as reference, the model would start generating a song, and when the output reached 97% completion, the model drew reference from a 3rd writer and 100% of the likeness in attribution came from writer 3.
So, does choosing between AI-generated alternatives make us the creator?
Even with highly specific prompting, the user cannot necessarily determine how the model translates those instructions into musical expression. The model may introduce musical elements, structures or stylistic characteristics that were never explicitly requested.
This raises an important distinction between creative intention and creative execution.
A user may know exactly what they want to hear, but that does not necessarily mean they have authored what the AI produces.
And importantly, the time spent prompting, refining and selecting an output does not automatically transform that process into copyrightable authorship.
The US Position: Human Influence Is Not Human Authorship
The US courts and Copyright Office have also addressed the distinction between human creativity and AI-generated expression.
In Thaler v. Perlmutter, the US Court of Appeals confirmed that copyright protection requires human authorship. The case concerned a work identified as autonomously generated by AI, rather than a work created through human prompting.
The US Copyright Office's January 2025 report on AI copyrightability went further in examining prompting. It concluded that prompting alone does not generally establish copyright in AI-generated outputs where the system, rather than the human, determines the expressive elements.
There is an important distinction here; Human influence is not necessarily human authorship.
Neither the German nor the US position establishes that AI-assisted works can never receive copyright protection. Rather, the question is whether the human contribution is sufficiently original and identifiable in the resulting output.
For music, that distinction is fundamental. A person who writes a melody, develops a composition and uses technology to record or arrange it has made identifiable creative contributions.
A person who asks an AI system to generate a song and subsequently selects their favourite version may not have done so.
Both have used technology. But have both created music in the copyright sense?
Isn't AI Just Another Creative Tool?
This is perhaps the most common argument in favour of generative AI.
We have always used technology to make music. From synthesisers, autotune and drum machines to Logic, Protools and other DAWs and digital sampling, technological development has continually changed how music is created.
And we agree. Technology itself does not undermine creativity.
Historically, these tools have generally allowed humans to determine the musical expression, whether by playing an instrument, programming notes, recording sounds, manipulating samples or arranging a composition.
A digital piano may reproduce sampled sounds from a real instrument, but the composer still determines which notes are played, when they are played and how they form a musical work.
The distinction is not whether the instrument is physical or digital. It is who determines the creative expression.
Generative AI complicates this relationship. Rather than simply executing musical decisions, a generative model may determine the melody, harmony, instrumentation, arrangement, vocal characteristics and structure itself.
Of course, AI functionality is increasingly being integrated into traditional music production software, and the distinction between technical assistance and autonomous generation is becoming less obvious.
This is why the question of human authorship matters more than ever.
We should not treat every use of AI as equivalent. There is a considerable difference between using AI to clean up a recording and asking it to compose and produce an entire song.
If There Is No Copyright, Where Is the Value?
This brings us to the next question.
If a substantial proportion of AI-generated music cannot establish sufficient human authorship to qualify for copyright protection, what happens to the enormous volume of AI music being uploaded to streaming platforms every day?And perhaps more importantly, where does the economic value lie?
An absence of copyright protection does not automatically prevent a recording from being distributed or monetised. Streaming platforms have their own policies, and other contractual or legal rights may apply.
However, the lack of copyright protection raises questions about ownership, exclusivity and long-term commercial value. We won’t get into the financial thievery of AI generated songs in the market share pools in this article.
If an uploader cannot claim exclusive copyright in the AI-generated musical expression, what exactly are they bringing to the market?
And how sustainable is a business model built around generating enormous quantities of music that may not enjoy the same legal protection as human-authored works? This becomes even more interesting when we consider consumer behaviour.
The Return of Human Creativity
During Musikförläggarnas Industry Day, we heard from music educators Musikmakarna about something rather encouraging and positively optimistic.
Students are taking pride in their own creative work. Writing their own lyrics, composing their own music and being able to demonstrate that their work was created without generative AI carries credibility. There is a growing distinction between making something yourself and generating something through a machine. And that form of credibility has value.
As transparency requirements develop and platforms introduce more visible AI disclosures, consumers may increasingly be able to distinguish between human-created and AI-generated music.This creates an interesting economic question.
When consumers are given a genuine choice, will they value music differently depending on how it was created?
We believe human creativity, artistic identity and authenticity will become increasingly important as commercial differentiators.
This does not mean that every listener will reject AI music. Nor does it mean that every human-created song will outperform an AI-generated alternative. But it does mean that the origin of a musical work could become a more significant part of its commercial value chain.
And if AI-generated music becomes less commercially attractive to uploaders, the incentives for flooding streaming platforms with it may begin to change.
Is the AI Music Business Model Sustainable?
Now consider the economics from the other side. AI music companies face increasing scrutiny over the copyright-protected material used to train their models.
Rights holders are demanding transparency, licensing arrangements and remuneration for the exploitation of their works.
We have previously discussed the importance of distinguishing between three potential value pools in AI music licensing:
Training: The use of copyright-protected works to develop AI models.
Prompting: The potential exploitation of identifiable protected works or other rights-relevant references during generation.
Output: The exploitation of the resulting AI-generated material, including potential infringement and licensing considerations.
What happens when AI companies must negotiate licences, address historical unlicensed uses, provide greater transparency and remunerate rights holders across multiple stages of exploitation?
At the same time, what happens if the commercial value of the resulting AI music decreases as consumers and platforms place greater emphasis on human creativity?
The cost of generating music could increase while the commercial value of distributing that music decreases.
Will the AI Music Bubble Burst?
Will AI music companies disappear? We doubt it.
Generative AI is a mind blowing technological development, and there is clearly a market for creative experimentation, entertainment and personal expression.
But we question whether the current business model, particularly the mass generation and commercial distribution of AI music, will remain sustainable in its present form.
We are not arguing that AI should disappear. Quite the opposite.
We believe there is room for AI to develop into a valuable, licensed and sustainable part of the music ecosystem. But the industry must distinguish between technological capability, human creativity and copyright ownership.
Until the legal boundaries around AI-generated music become clearer, and the economics of licensing and distribution begin to reflect those boundaries, the current model remains unstable.
Perhaps the bubble will burst. Or perhaps the industry will simply be forced to evolve.
Either way, we remain optimistic about the future of music.