Deezer: Music AI Threatens Artists' Earnings

A Proliferation of AI-Generated Tracks

On the Deezer platform, around 75,000 tracks created by artificial intelligence are added each day, accounting for nearly 44% of new content. Although this figure has increased sevenfold in just over a year, these tracks represent only between 1% and 3% of total listens. This imbalance between massive production and limited attention may seem trivial for artists, but the rise of AI-generated music is profoundly transforming the music offering, jeopardizing the revenue distribution model and the visibility of existing works.

The Disappearance of Barriers to Entry

The immediate impact of this dynamic is the transformation of the music supply structure. The barrier to entry, already low in streaming, is tending to disappear. Creating, testing, and distributing a track has become an almost instantaneous and low-cost operation. The growth of the catalog is no longer hindered by production costs but is fueled by increased technical capacity, freed from any economic constraints. This abundance makes capturing attention more difficult in a continuous stream of new tracks.

A Distribution Model Under Pressure

Contrary to the common belief that each artist is paid based on the listens of their own tracks, the dominant economic model of streaming, known as pro-rata, relies on pooling. Subscription and advertising revenues are aggregated and redistributed according to each artist's share of streams. This system, used by Spotify, Apple Music, and Amazon Music, is being tested by the multiplication of available tracks, which fragments listening and erodes unit compensation.

Deezer adopted a user-centric model in 2024, where each listener's subscription is redistributed only to the artists they have listened to. This system reduces the dilution effect by restoring a direct link between individual listening and compensation, but it does not solve all the problems. The saturation of the catalog harms the discoverability of works, and listening fraud remains a threat.

The Challenges of Algorithmic Recommendation

The pressure on recommendation algorithms is another crucial issue. With the expansion of the catalog, the automated prioritization of content becomes uncertain. Recommendation systems, designed to connect an abundant supply to individual preferences, are facing increasing background noise. The dissemination of micro-listens across a long tail of tracks contributes to redistributing revenues at the expense of established works.

Moreover, a notable proportion of listens to AI-generated content is classified as fraudulent by Deezer and is subject to demonetization, highlighting that the problem goes beyond mere music production and touches on the integrity of the distribution model.

AI as an Economic Arbitration Tool

The ability to produce massive volumes of content at almost no cost paves the way for unprecedented optimization strategies. The stakes become decidedly economic, with some players able to exploit compensation mechanisms by multiplying tracks, algorithmically assessing their performance, and artificially inflating listen counts.

A Migration of Value

Beyond issues of fraud, the most striking evolution is the gradual erasure of scarcity. In an environment where music production can be automated on a large scale, the value of content tends to shift. Some music, particularly functional music, lends itself to automated generation, while others, based on artistic uniqueness, retain a unique value.

The Crucial Role of Platforms

In this context, streaming platforms play an arbitrating role. Deezer has implemented several initiatives to contain the effects of this transformation: identifying AI-generated content, excluding these tracks from recommendations, demonetizing fraudulent listens, and developing detection technology offered under license to the industry.

Deezer's CEO, Alexis Lanternier, emphasizes that generative music AI is no longer a mere epiphenomenon and that the ecosystem must mobilize to defend creators' rights and ensure transparency towards listeners.

An Issue Beyond Platforms

The implications of this transformation extend beyond streaming services. A study with CISAC mentions a risk to a significant share of creators' revenues by 2028. Although the extent of this impact is uncertain, this estimate underscores the vulnerability of the current model in the face of rapidly changing production conditions.

At the same time, user expectations are evolving. A majority of listeners want AI-generated content to be clearly identified, making transparency a key element of the relationship between platforms, creators, and the public.

An Ongoing Restructuring

The rise of AI-generated music does not yet mean the displacement of artists, but reveals a deeper tension related to abundance and the redistribution of value. In an environment where production is no longer limited by scarcity, the ability to capture attention, organize distribution, and preserve uniqueness becomes crucial. The question extends beyond musical creation to encompass the conditions of exposure, valuation, and compensation for this creation.

The trajectory remains open, and we are only at the beginning of generative AI. It will be shaped by platform decisions, the evolution of the regulatory framework, and the ability of the economic model to absorb a potentially devastating wave. One thing is certain: the main issue is no longer to produce more music, but to preserve the conditions of its value.