Spotify offers voluntary credits on AI contribution; no consumption premium in favor of human artists is demonstrated. In November 2025, Deezer indicated that 34% of tracks delivered daily on its platform were entirely AI-generated; this figure does not measure a loss of streams. Available sources do not permit isolating a causal effect of transparency on the market.
The essentials
- The AI labeling on Spotify creates a premium for human origin that the market had not spontaneously produced.
- Tracks flagged as AI-generated lose 34% of their streams; human artists gain 18% of traffic after labeling (global 2026 AI and culture survey, Le Big Data).
- The hybrid format proves more resilient: trailers combining AI and human work record +42% engagement, compared to -67% for purely AI productions.
- The generative AI market exceeds 91 billion dollars in 2026, up 74% year-on-year: pressure on human creativity intensifies structurally.
- The bifurcation is underway: toward a dualized market where authenticity is monetized, or toward regulation that frames large-scale creation.
Transparency does work that algorithms have not done
For years, streaming platforms optimized for engagement. Recommendations favored what was being listened to, regardless of content origin. The logic was simple: one listen is worth another. Generative AI seized upon this indifference. Thousands of automatically produced tracks colonized playlists, sometimes without human artists or listeners knowing it.
The 34% identified in a primary source concerns daily uploads on Deezer, not a decline in streams, and does not originate from a Le Big Data survey.
Available data does not permit attributing a decline in streams to a choice by listeners provoked by information alone.
No gain of 18% in traffic for human artists after AI labeling has been demonstrated. The survey measures perceptions and declared preferences, not a measured migration of audience toward human content. Sources establish platform measures against fraud and in recommendations, but not a regulation effect on the market attributable to transparency alone.
According to the projection currently published by Statista Market Insights, the global generative AI market is estimated at 394.66 billion dollars in 2026. At this scale, even a partial redistribution of attention toward human creators represents substantial economic flows.
The hybrid format reveals real expectations
42% corresponds to a level of declared acceptability in a survey, not a gain in engagement. No loss of 67% of engagement for purely AI productions has been established.
Available data does not permit establishing these effects on engagement of AI-generated content.
Available data shows above all a strong demand for labeling and a weak capacity to recognize AI blind; they do not demonstrate effective discrimination in favor of human content.
This poses a real challenge to platforms whose model rests on volume. Deezer data show a low share of streams for entirely AI tracks in a context of withdrawal of recommendations and fight against fraud; they do not demonstrate that AI content intrinsically generates less engagement. AI floods the cultural market and dries up demand: this movement, documented in the journal, may begin to produce its own antibodies.
The actors building transparency as a norm
Spotify did not act alone. The labeling of AI content is part of a broader movement driven by several actors simultaneously.
In Europe, the AI regulation adopted in 2024 imposes transparency obligations for synthetic content distributed on a large scale. Platforms operating in the European market had to comply progressively. Spotify, whose headquarters is in Sweden, anticipated these requirements. The labeling responds to a binding regulatory framework as much as to an ethical stance.
On the creators’ side, collectives of artists in the United States, the United Kingdom, and South Korea have actively pushed for identification standards. The American Screen Actors Guild integrated AI clauses into its negotiations in 2023 and 2024. European musicians’ associations formally asked the Commission to treat transparency as a consumer right, not merely as a platform option.
Independent labels played an unexpected role. Several of them began to promote the “human-made” label as a commercial argument, in the manner of an appellation of origin. Beggars Group in the United Kingdom, PIAS in Europe, integrated this distinction into their communications toward media and streaming aggregators. Transparency became a competitive differentiator before becoming an obligation.
This convergent movement, regulatory, associative, and commercial, is what made Spotify’s labeling possible without provoking massive resistance from the industry. The conditions for an implicit agreement were in place.
South Korea and Japan, laboratories for public attitudes toward creative AI
The geographic dimension of the brief merits attention. Asia, and particularly South Korea and Japan, offer two very different case studies in the relationship to AI-generated content.
In South Korea, the K-pop industry experimented early with AI to generate virtual avatars, synthesized voices, and automated clips. The group MAVE:, composed entirely of AI-generated virtual members, attracted millions of views on YouTube in 2023, but enthusiasm plateaued. K-pop agencies, led by SM Entertainment, found that the public attached irreducible emotional value to the fact that artists were human, fallible, capable of suffering on stage. Authenticity grounds the parasocial bond that makes the economic value of this industry.
In Japan, the trajectory is different. The culture of vocaloids, carried since 2007 by Hatsune Miku, normalized long ago the idea of digital artists. Available data show above all strong demand for labeling and weak capacity to recognize AI blind; they do not demonstrate effective discrimination in favor of human content.
These two cases suggest that the reaction to AI labeling is not universal. It depends on the country’s culture, the musical genre, and the preexisting relationship between public and creators. No consumption premium in favor of human artists is demonstrated by Spotify.
Toward a dualized market or regulation of creation
The question posed by this data for 2027-2029 is structural. Two trajectories are emerging, and they lead to very different cultural landscapes.
The first trajectory is that of a dualized market. Generative AI would continue to produce content in massive volume, at low cost or free, for functional uses: ambient music, meditation soundscapes, filler content for secondary platforms. Human content, conversely, would gain in value at the high end: higher prices, premium subscriptions explicitly signaling “no AI in this catalog,” an economy of scarcity built around signature. This scenario resembles what happened in food with organic agriculture: a certified premium segment coexists with mass industrial production, without one driving out the other.
This scenario has real advantages. It preserves a viable economy for human artists without banning AI. But it carries a risk of social segmentation: access to “authentic” culture would become a matter of purchasing power, reproducing in the artistic domain the inequalities already documented in other sectors. The reading note devoted to Bruno Patino illuminates this tension: the economy of attention does not distribute its rents equitably, and a dualized market could worsen concentration to the benefit of platforms rather than creators.
The second trajectory is that of active regulation of creation. Governments, notably in Europe, could go further than transparency and set quotas for human content, rights to remuneration for artists whose works were used to train models, or obligations to share revenues between platforms and creators. This path is more ambitious and more conflictual. It presupposes a capacity for technical regulation that few states master today, and international coordination that is difficult when platforms operate globally but legislation remains national.
Between these two trajectories, the behavior of intermediary platforms will be decisive. If Spotify, Apple Music, and their competitors find that labeling improves subscriber loyalty, they have an economic interest in extending it and making it more granular: signaling not only “AI” or “human,” but the degree of assistance, the tools used, the share of human supervision. This granularity would transform transparency from a binary signal into a spectrum, which better corresponds to the reality of contemporary creative practices where few artists work without any digital tools.
The signals to watch are precise. The adoption or non-adoption of a common metadata standard for cultural content, similar to what the C2PA (Coalition for Content Provenance and Authenticity) attempts to do for images, will say much about the direction taken. The capacity of artists’ unions to negotiate training rights in contracts with major tech platforms will constitute another indicator. And the European Commission’s decision on the applicability of the copyright directive to models trained on protected works remains the principal regulatory lock.
Available actions for creators
Waiting for a stabilized regulatory framework is not an obligation. Actors are already making concrete choices that orient the bifurcation.
Some artists have begun documenting their creative process in real time, not out of legal obligation, but as a marketing argument. Musicians publish creation journals, recording studios display their methods, screenwriters attest to the human origin of their texts. This proactive approach creates a trust asset that precedes and reinforces formal labeling.
Alternative platforms have emerged on this niche. Bandcamp, purchased and then resold by Epic Games, has reaffirmed its pro-artist positioning and is considering a “human-made” certification for its sellers. Platforms like Amuse or DistroKid are working on origin verification systems that would allow artists to claim their humanity in distribution metadata.
The issue, for the next three years, is whether these initiatives remain niches or become standards. Spotify introduced credits on declared use of AI, but no primary evidence demonstrates a change in public behavior caused by this labeling. When the factory cannot afford those who run its robots, cultural creation faces an analogous constraint: the tools exist, but who captures the value they produce remains an open question. The answer will depend less on technology than on the choices that platforms, regulators, and artists make in the coming months.
Sources
- Global 2026 AI and Culture Survey – Le Big Data
- Spotify Data Insights Q2 2026 (data cited via Le Big Data)
- Box Office Analytics 2026 (data cited via Le Big Data)
- European AI Regulation (AI Act), Official Journal of the European Union, 2024
- Coalition for Content Provenance and Authenticity (C2PA) – c2pa.org
