Deezer Has an AI Music Problem — But Who Created the Streaming Economy That Made It Profitable?

Deezer increasingly presents the explosion of AI-generated music as one of the great threats facing the modern streaming industry, and there is no question that the numbers are staggering. The company has reported AI-generated music arriving at rates approaching 90,000 tracks per day at peak levels, with fully AI-generated recordings accounting for more than half of new daily uploads at those levels.

Deezer also says it detected and tagged more than 13.4 million AI-generated tracks during 2025. Those figures are extraordinary, but focusing exclusively on the technology allows the streaming industry to avoid a much more uncomfortable question: why was the music business already structured so perfectly for this kind of industrial-scale content production? Deezer did not invent generative AI, and it would be dishonest to blame one company for creating the entire streaming economy, but Deezer is an important participant in the system that turned music into endlessly scalable digital inventory.

The industry built the distribution infrastructure, standardised the metadata, normalised subscription access, measured consumption through streams and created an environment in which recordings could be uploaded and processed at enormous scale. AI arrived and discovered that the system could be exploited even more efficiently than human beings had ever imagined.

The distinction matters because criticism becomes meaningless when it turns into exaggeration. Deezer is not responsible for every weakness in streaming, nor did it invent the concept of digital distribution. But Deezer is a corporation operating inside an ecosystem whose economic incentives are increasingly detached from the human labour traditionally required to create music. The platform earns revenue by providing access to an enormous catalogue and by maintaining a listening environment capable of keeping subscribers engaged.

That creates a structural appetite for content. More recordings mean more potential discovery, more playlist material, more recommendation possibilities and more reasons for listeners to remain inside the service. For years, that abundance was treated as one of streaming’s greatest achievements. Now the industry is discovering the darker side of abundance: when the cost of producing another recording collapses, there is nothing stopping the catalogue from being flooded with material that has little or no genuine audience behind it.

The modern streaming economy did something revolutionary to music. It removed much of the physical scarcity that once constrained the industry. A record label once needed manufacturing, warehouses, physical distribution and retail space to move products. Digital streaming eliminated most of those limitations. A recording could exist as data, sit inside a platform catalogue and become available almost anywhere with an internet connection. That transformation was enormously convenient for consumers and opened opportunities for independent artists, but it also created a new economic reality in which the cost of adding another recording to a digital catalogue became dramatically lower.

Once distribution becomes frictionless, production becomes the next bottleneck. Generative AI attacks that bottleneck directly. It does not need to create a new streaming service. It simply produces more material for the existing ones.

The speed of the change should terrify anyone who cares about the future of human-made music. Deezer reported that in January 2025 it was receiving approximately 10,000 fully AI-generated tracks every day, representing around 10% of daily deliveries. By April, the company said that number had climbed beyond 20,000 tracks per day and represented more than 18% of new uploads. By September, Deezer reported more than 30,000 AI-generated tracks arriving every day, representing over 28% of daily deliveries. In early 2026, the figures continued climbing, with Deezer reporting approximately 60,000 AI-generated tracks per day in January and around 75,000 per day by April. By June, the company reported peak levels of roughly 90,000 AI-generated tracks per day, exceeding half of all new deliveries. This is not a marginal technology entering an established marketplace. It is a production system capable of overwhelming the catalogue itself.

The more revealing part of Deezer’s own data is that AI-generated music represents a much smaller share of actual listening than it does of new uploads. Deezer has reported that fully AI-generated music accounts for approximately 1–3% of total streams, while also saying that up to 85% of streams associated with fully AI-generated tracks were fraudulent in 2025.

That tells us that the AI crisis is not primarily a story about consumers abandoning human musicians for machines. It is a story about economic incentives. If a person can generate enormous quantities of music at minimal cost, the temptation is to exploit whatever mechanism turns streams into money. The recording becomes a vehicle for attempting to generate economic activity. The music itself can become secondary. That is precisely where the streaming model becomes vulnerable: the system has a measurable unit of value, and technology has made it possible to manufacture an enormous number of objects capable of participating in that system.

This is where the corporate narrative becomes uncomfortable. It is easy for a streaming platform to point toward AI and describe the problem as something malicious actors are doing to the industry. There is truth in that description, particularly when fraudulent streaming is involved. But it is incomplete. The people exploiting the system did not invent the idea that a stream has economic value. They did not invent algorithmic recommendation.

They did not invent the obsession with catalogue scale. They did not invent the transformation of music consumption into behavioural data. They are exploiting an environment that the digital music industry spent years constructing. Deezer is now deploying increasingly sophisticated detection systems because that environment has become difficult to police. The platform is essentially trying to defend an industrialised streaming economy from industrialised content production.

The underlying contradiction is almost impossible to miss. The streaming industry spent years celebrating the fact that it could make enormous catalogues available to consumers. Platforms competed on breadth, convenience and discovery. The promise was that the listener could access practically everything. But there is a fundamental difference between making existing human-created music accessible and creating an economic system where the quantity of new recordings can expand without meaningful physical limitations.

Human creativity is constrained by time. AI-generated content is constrained primarily by computing resources, distribution policies and economic incentives. Once those constraints diverge, the catalogue can expand much faster than human attention can possibly follow.

That is the real crisis.

There is no shortage of music.

There is a shortage of human attention.

Deezer says that detected AI-generated tracks are labelled and that fully AI-generated recordings are excluded from algorithmic recommendations and editorial playlists. It also says fraudulent streams are removed from royalty calculations. These policies may help protect legitimate artists from manipulation, but they reveal something far more important about the modern music business: Deezer is not merely a pipe carrying music from artist to listener.

The company increasingly acts as a gatekeeper over discovery, visibility, classification and economic participation. The platform decides what is shown, what is recommended, what is excluded and which streams count. That is enormous cultural authority for a private company.

When Deezer places an AI-generated label on a recording, the platform is making a classification that can affect how a listener perceives the work. When it excludes a track from recommendations, the platform is making an even more consequential decision because it is altering the recording’s opportunity to be discovered. When fraudulent streams are removed from royalty calculations, the platform is determining which consumption events are economically legitimate.

None of these actions is automatically wrong. The problem is the concentration of decision-making power. Artists are increasingly dependent upon automated systems operated by corporations that can change their policies without artists having anything resembling an ownership stake in the infrastructure.

That is not the same thing as independence.

It is access granted under corporate conditions.

There is another development artists should watch carefully: Deezer has increasingly positioned its AI detection technology as something that can be used beyond its own platform. The company has announced partnerships involving the technology and has indicated that its detection capabilities can serve wider music-industry needs.

Deezer has also discussed patent applications connected to its detection technology. Commercialising technology is not inherently suspicious, but the development demonstrates how a platform can turn a problem inside its ecosystem into a new commercial opportunity. The same corporation operating a streaming service can become a technology provider helping other organisations decide whether music was generated using AI. That makes the questions around transparency, accuracy, appeals and accountability more important, not less.

The absurdity of the situation becomes clearer when you step back. AI can generate music. Streaming platforms distribute it. Algorithms recommend it. Other algorithms detect it. Fraud systems analyse its streams. Classification systems label it. Recommendation systems suppress or promote it. The artist is now sitting in the middle of a machine-to-machine economy while being told that the system exists to serve creativity. At some point, the industry has to admit that the technological infrastructure has become so dominant that the human creator is no longer necessarily the central participant in the process.

The artist provides the cultural origin, but machines increasingly control distribution, measurement, classification and discovery.That should concern every musician, regardless of whether they use AI.

Technology has always been part of music. Drum machines, synthesisers, samplers, digital audio workstations, pitch correction, software instruments and automated mastering have all transformed the creative process. The important distinction is not whether technology touches a recording. The important distinction is who is exercising creative control and how the economic system treats the resulting work. A human artist using technology to create something intentional is not necessarily equivalent to an automated system producing thousands of tracks designed primarily to participate in a financial system.

Treating those situations as identical would be intellectually lazy. The industry needs more precise language and more nuanced policies than simply dividing music into “human” and “AI” categories.

The streaming economy was built around an assumption that music would remain expensive enough to produce that catalogue growth would retain some relationship to human labour. AI destroys that assumption. Once a recording can be produced with minimal labour and almost no physical distribution cost, scarcity disappears from production. That changes the economics completely. The value of a recording can no longer be protected simply by the fact that producing another recording requires another human being to spend weeks or months doing it. The marketplace can now be flooded.

That means the future value of music increasingly depends on things machines cannot manufacture at infinite scale: identity, trust, reputation, community, authenticity, cultural context and direct relationships between artists and listeners.

For an independent musician, the situation is particularly brutal because the artist is already competing against enormous corporate catalogues. Now they are potentially competing against automated production systems capable of creating thousands of tracks in the time it takes one human being to finish a serious record. The traditional advice to “release more” becomes almost meaningless under these conditions. How does a human compete with infinite output? The answer is that they cannot and should not attempt to. The independent artist’s competitive advantage is not quantity. It is personality, history, craftsmanship, identity and connection. The danger is that platforms built around scale may not always reward those qualities as strongly as they deserve.

A platform can measure that a song was played. It can measure how long somebody listened. It can measure whether somebody skipped it. It can measure whether somebody saved it. It can measure whether listeners returned. But none of those measurements fully explains why a piece of music matters to a human being. A song can have modest numbers and enormous personal significance. An underground record can become foundational to a small community without ever approaching mainstream streaming statistics. A track can matter because it represents a place, a friendship, a trauma, an era or a belief. Those meanings do not fit neatly into a dashboard.

Yet the modern music industry increasingly governs itself through dashboards.

That is where the cultural rot begins.

Before AI-generated music became a serious commercial problem, musicians were already being encouraged to reshape their behaviour around platform mechanics. They were told to release frequently, optimise metadata, chase playlists, create short-form content, monitor engagement and study analytics. The artist increasingly became responsible not only for making music but for feeding the platform’s content ecosystem. The system rewarded those who understood its mechanics. It encouraged musicians to think about how a song would perform before they finished making it.

Then AI arrived and took that logic literally.

If the platform wants constant content, generate constant content.

If the platform rewards streams, pursue streams.

If the platform values engagement, manufacture engagement.

The machine simply removed the human limitations from a strategy the industry had already normalised.

The platform now has to distinguish genuine cultural production from industrial content production, but that distinction is extremely difficult because the streaming economy treats both as catalogue entries. A masterpiece and a disposable synthetic track can both arrive as audio files with metadata. The platform can classify them, but the underlying system initially treats them as structurally similar objects. This is one of the deepest contradictions in digital music: culture is human, but the infrastructure increasingly processes culture as data.

AI has simply made that contradiction impossible to ignore.

Streaming is incredibly convenient, but convenience has consequences. When music becomes an infinite subscription catalogue, the psychological relationship between listener and artist can change. Buying a record was an act of commitment. Downloading an album could still feel like acquiring something. A direct purchase creates a relationship between consumer and creator. Streaming turns that interaction into access. The listener does not necessarily own anything. The artist does not necessarily receive a meaningful direct relationship with the listener. The platform sits between both parties and monetises the connection.

That arrangement works brilliantly for scale.

It is less obviously brilliant for independence.

There is a strange dual role emerging in the streaming business. Platforms want to be marketplaces with enormous catalogues, but they also want to become authorities capable of deciding what belongs in those catalogues, what deserves recommendation and what constitutes suspicious activity. Deezer’s AI strategy demonstrates that tension clearly. The company wants the catalogue to remain open enough to support scale while simultaneously policing that catalogue with increasingly powerful automated systems. It wants to be both the highway and the traffic police.

The danger is that artists increasingly have no choice but to accept both roles.

This is why direct ownership matters more now than ever. Artists should not treat streaming profiles as their permanent homes. They are useful storefronts and discovery tools, but they are controlled environments. A platform can alter its policies, change its algorithms, introduce new labels, change its business model or restructure its economics. The artist may have little meaningful influence over any of it. Building a direct audience is therefore not a nostalgic return to the past. It is a modern survival strategy.

The artist should maintain their own archive, protect their rights documentation, keep direct contact with listeners, maintain an independent web presence and provide somewhere audiences can support them without another corporation standing between them and the transaction. The point is not to abandon streaming. The point is to stop pretending streaming is ownership.

UNIDARK is an example of why that distinction matters. The project maintains a broader independent ecosystem rather than allowing one platform to become the sole gateway between the music and the audience. The Official UNIDARK Hub provides a central destination for releases, videos, lyrics, streaming platforms and project information, while the Official UNIDARK Store creates a direct route for listeners who want to support the work. That model does not require pretending that streaming platforms have no value. It simply refuses to give them absolute control over the relationship.

That is the difference between using a platform and depending upon a platform.

Artists should also remember that Deezer is a business. Its interests are not identical to theirs. Deezer reported €267.1 million in revenue during the first half of 2025 and highlighted its progress toward profitability, cost discipline and commercial growth. By the first half of 2026, the company reported €268 million in revenue and continued profitability while simultaneously highlighting the rapidly expanding AI-music problem. (newsroom-deezer.com) None of this proves misconduct. It does, however, provide essential context. When a corporation discusses protecting artists, fighting fraud and improving the music ecosystem, artists should also remember that the corporation has shareholders, employees, operating costs, commercial objectives and a responsibility to generate revenue.

Artists have different objectives.

The artist wants the work to survive.

The corporation wants the business to survive.

Those goals can overlap without being identical.

The music industry is full of phrases such as “artist-centric,” “artist-friendly,” “fairer remuneration,” “better discovery” and “protecting creators.” Some of these initiatives may genuinely benefit musicians. But artists should judge platforms by outcomes, contracts, policies, transparency and control rather than marketing language. A company can sincerely claim to support artists while simultaneously building a business model that makes those artists increasingly dependent on its infrastructure.

Good intentions do not eliminate structural power.

The more Deezer expands its role in determining what is AI-generated, the more important it becomes to ask how artists can challenge those decisions. A classification can influence how listeners perceive a recording. Recommendation exclusions can affect discovery. Fraud detection can affect economic participation. If an artist believes a platform has misclassified their work, the process for challenging that decision matters enormously. Automated judgement without meaningful human review can become a new form of gatekeeping, particularly when the platform is simultaneously the distributor, recommendation engine and arbiter.

The industry should not replace old gatekeepers with invisible ones and then call the result progress.

Listeners have a role in this debate because their behaviour determines which parts of the ecosystem become economically powerful. Consumers should ask whether the music service they use actually helps them discover artists or simply keeps them inside an engagement system. They should ask how much of their subscription reaches creators, how platforms handle fraudulent activity and what rights they have when content is removed or reclassified. Most importantly, listeners should recognise that direct support has a different economic meaning from passive consumption.

When you buy directly from an artist, the artist knows you chose them.

When you stream inside a platform, the platform controls the relationship.

That distinction is becoming more important as digital music becomes increasingly automated.

The underlying battle is therefore not really “human music versus AI music.” It is a battle over scarcity, attention and economic value. Human creativity remains scarce because human time is scarce. AI-generated content is abundant because machine production can scale rapidly. Streaming platforms are caught between those two realities. They need enormous catalogues to provide value to subscribers, but they also need to prevent catalogue abundance from destroying discovery and economic integrity.

That is an almost impossible balancing act.

And artists should not be expected to solve it for them.

The reason independent artists still matter is precisely because human creativity contains something industrial production struggles to reproduce: a real relationship between creator, work and audience. The history of an artist matters. The circumstances surrounding a recording matter. The imperfections can matter. The struggle can matter. The community surrounding a genre can matter. A record can become important because people know who made it, why it was made and what it represents.

That is not efficiently scalable.

And perhaps that is exactly why it is valuable.

The lesson is not that artists should panic and remove themselves from every streaming service. That would be unrealistic and potentially counterproductive. The lesson is that artists should stop treating platforms as permanent homes. A streaming profile is an outpost. Your catalogue is an asset. Your audience relationship is an asset. Your direct store is an asset. Your website is an asset. Your mailing list is an asset. Your rights documentation is an asset.

The platform should be one part of the ecosystem, not the ecosystem itself.

This is the central contradiction that Deezer cannot escape. The streaming economy wanted scale, so it built systems capable of handling extraordinary catalogue growth. It wanted convenience, so it removed friction from distribution. It wanted measurable consumption, so it turned listening into data. It wanted enormous choice, so it expanded catalogues relentlessly. It wanted algorithmic discovery, so it gave machines increasing influence over what listeners encounter.

Then AI arrived and exploited every one of those characteristics simultaneously.

The industry wanted more content.

AI provided it.

The industry wanted more scale.

AI provided it.

The industry wanted frictionless distribution.

AI provided it.

The industry wanted measurable consumption.

AI provided a target.

The industry wanted catalogue abundance.

AI made abundance almost meaningless.

That is the rot underneath the current crisis.

The temptation will be to respond by giving platforms even more power. More detection. More classification. More surveillance. More automated moderation. More centralised databases. More rules. More labels. More systems deciding what artists can and cannot do.

Some of those tools will be necessary.

But if every solution increases the platform’s authority over artists, the cure can eventually become another disease.

The music industry should be careful not to fight algorithmic exploitation by creating an even more powerful algorithmic gatekeeper.

A musician can have millions of streams and still have very little control over their career. A platform can make an artist visible without making that artist independent. A playlist placement can generate enormous exposure without creating a direct audience relationship. A streaming profile can look like a business while functioning primarily as a rented storefront.

That distinction is essential.

Visibility is useful.

Ownership is power.

The strongest independent artists will not necessarily be the ones who reject technology. They will be the ones who understand technology without surrendering control to it. They will use streaming for discovery, social platforms for communication, distributors for logistics and digital tools for production, while maintaining direct ownership over the foundations of their work and their relationship with listeners.

That is the model worth defending.

Not isolation.

Not nostalgia.

Independence.

The most important lesson from Deezer’s AI problem is not that AI is dangerous, although certain forms of AI-generated catalogue manipulation clearly are. The deeper lesson is that artists become vulnerable when the infrastructure connecting them to audiences is controlled by companies whose priorities can change independently of the creators who supply the content. The artist who depends entirely on one platform is exposed to every policy change, algorithm change, ownership change, moderation decision and commercial restructuring that platform makes.

The artist who has built direct alternatives has leverage.

That leverage is freedom.

The streaming era has encouraged musicians to think of their careers through platform statistics. AI threatens to make that mindset even more destructive by introducing competitors capable of generating content at industrial speed. The answer is not to become more obsessed with numbers. It is to become more obsessed with ownership, identity and direct connection.

A human artist does not need to produce more music than a machine.

They need to create music that means something to someone.

And they need to make sure the relationship with that person does not belong entirely to a corporation.

Deezer can improve detection. It can label AI music. It can remove fraudulent streams. It can adjust recommendations. It can license detection technology. It can build increasingly sophisticated systems for policing the catalogue. Those measures may be necessary, and some may genuinely protect artists. But none of them changes the fundamental contradiction created by streaming: music has been transformed into infinitely scalable digital inventory, and now technology has arrived that can exploit that scalability far beyond the limits of human production.

The industry therefore needs to stop pretending that AI is simply an outside enemy attacking an otherwise healthy ecosystem. AI is exposing the ecosystem’s weaknesses. It is showing what happens when an industry treats content as infinitely reproducible, attention as the scarce resource and streams as a primary economic measurement.

Deezer did not create all of those conditions.

But Deezer operates inside them, profits from them and now has to defend them.

That is why artists should look beyond the company’s AI headlines and examine the structure underneath.

The real question is not whether Deezer can detect AI music.

The real question is whether the streaming economy can continue treating music primarily as scalable content without destroying the economic and cultural conditions that allow human artists to matter.

And if the answer is no, then the next revolution in music will not simply be about artificial intelligence.

It will be about artists taking back control.

UNIDARK, also known as Morning Star, is a UK-based independent extreme metal producer and the creator of Blackdeathgrin Metal — an original extreme metal genre combining elements of black metal, death metal, deathcore, and grindcore.

Explore the Official UNIDARK Hub

The central directory for music, releases, streaming platforms, videos, lyrics, Blackdeathgrin Metal information, and all official project links.
→ Official UNIDARK Hub

Explore Official Lyrics & Song Meanings

Discover the complete UNIDARK lyrics archive, song interpretations, and the concepts behind the music.
→ Lyrics Archive

Watch Official UNIDARK Music Videos

Experience the visual side of the project through official releases and video content.
→ Video Archive

Explore Metal Insights

Read more deep articles covering extreme metal genres, underground music, industry topics, production, platforms, culture, and the evolution of heavy music.
→ Metal Insights

Explore Sync Licensing

Read more deep articles covering music licensing, sync rights, music for film, television, games, advertising, trailers, independent production, hybrid music production, catalogue clearance, and the evolving relationship between music and visual media.
→ Sync Licensing

Explore the UNIDARK FAQ Archive

Find answers about UNIDARK, Morning Star, Blackdeathgrin Metal, releases, the creative process, and the philosophy behind the project.
→ UNIDARK FAQ

Support UNIDARK Directly

Support independent extreme metal creation through the official UNIDARK store.
→ Official UNIDARK Store

Join the Official UNIDARK Transmission

Subscribe for new releases, Metal Insights articles, official announcements and updates directly from UNIDARK. Follow the evolution of Blackdeathgrin Metal and the continuing development of the UNIDARK project.