All articles

Deezer and TIDAL AI Music Royalties Explained

Two popular streaming platforms have made decisions to deny machine-created music the right to earn royalties the way human-made music does, and they have done it in two completely different ways. If you create music yourself or help artists with distribution of their tracks and want to know how the process works and how much of your subscription fee ends up on the creators’ accounts, then the issue of Deezer and TIDAL AI royalties is the most important change in streaming industry that I saw since the 1,000-stream threshold was introduced by Spotify.

Before we move into details, here is the summary. TIDAL announced its new AI policy on June 29, 2026, and stated that it will never knowingly pay royalties on wholly machine-generated tracks, with AI tags and fraud prevention features becoming effective on July 15. Deezer has another strategy to deal with AI music: the platform detects and tags such tracks, excludes them from algorithmic recommendations and editorial playlists, and demonetizes the fraud associated with them. In practice, this means that almost 85 percent of all AI music streams on Deezer are affected. The first platform cuts royalties on the track level, and the other one does it on the stream level. These distinctions sound theoretical until you have your money on it, and I am going to explain exactly why later.

For the past couple of years I have been helping independent artists manage their distribution and royalty reports, and in 95 percent of cases this topic is discussed now. So, here is what each platform actually does, how it affects the money flow and the place where, in my opinion, common pieces of advice you find on the Internet are wrong.

Why Do Platforms Care About Royalties for AI Music

The scale of the issue is truly hard to underestimate. As stated in Deezer’s newsroom announcement from July 2026, currently, the platform receives about 90,000 tracks that were created entirely with the help of artificial intelligence on a daily basis, with the peak day bringing more than half of all the uploads. Just six months ago, in January 2025, AI-generated music accounted for 10 percent of all uploads, by June 2025 it was 18 percent, by April 2026 it was 44 percent, and now it has exceeded 50 percent. No human A&R department in the history of the genre has ever faced such a flood.

The logical question here is: why would anybody upload tens of thousands of tracks that nobody needs. The answer is pretty simple and it is related to the money, with streaming royalties being the only ones involved in this process. As you know, royalties come from a shared pool supported by subscriptions and ads. Every stream, whether it is real or fake, takes a share of the pool. The more tracks you upload and the more bots you hire, the easier it is for you to get some money from working musicians without making any effort.

That is not a hypothetical scenario, it happened already. In March 2026, Michael Smith, a man from North Carolina, pled guilty to wire fraud conspiracy of receiving more than 10 million dollars in royalties between 2017 and 2024. He had bought hundreds of thousands of AI-generated tracks, uploaded them to Spotify, Apple Music, Amazon Music, and YouTube Music, and was using bots that were streaming them for billions of times per month. Once he admitted that his bots could generate about 661,440 streams per day. He agreed to forfeit 8,091,843 dollars and 64 cents, and the authorities called this case the first one of its kind. It definitely will not be the last.

At least Deezer is able to give the figures of this activity on its own platform. The company discovered that up to 85 percent of streams on fully AI-generated tracks in 2025 were fraudulent. Read that once again. Most of plays on the music that was generated with the help of artificial intelligence were never real listeners at all. After you understand this fact, both platforms’ policies start looking not like some kind of technophobic declarations, but like the necessary maintenance of the pool.

How Does Deezer Handle AI Music Royalties

Deezer was the first platform to introduce an AI policy, and, in my opinion, it has the most advanced system to deal with the problem. The platform developed its own AI detection system that was trained on 94 million songs, and then it filed two patents on this technology in December 2024, which were published in June 2026 both in EU and US patent offices. The system works automatically on the stage of ingestion, which means that every track is checked for AI generation the moment it comes to the service.

When a track is identified as fully AI-generated, three things happen. First, it gets the visible AI tag. Deezer was the first streaming platform to label the music of this kind with a special tag that was visible to listeners, and this practice started in June 2025, and for a long time it was the only one. Second, the track is excluded from algorithmic recommendations and editorial playlists. It is still available in the library of the service and is searchable through search and artist page, but it will never be recommended by Deezer. Third, all the streams on the track that are identified as fraudulent are demonetized and excluded from the pool of payments entirely.

Now comes the point that every summary of the platform’s policy misses in one way or another, so pay attention to this piece of information. Deezer does not demonetize all the AI tracks automatically. A listener who searched for a track generated with the help of AI and streamed it can still receive a royalty on it, at least according to the current version of the policy. What the company excludes from the pool is the fraud, and, according to the company’s statistics, fully AI-generated music accounts for only 1 to 3 percent of all the streams on the platform, and, with up to 85 percent of those streams demonetized as fraudulent, the impact on the pool is minimal. Deezer’s own documents confirm this, saying that the company provides the classification, and each operator decides on its royalty.

Deezer stopped storing hi-res audio versions of AI tracks in April 2026, which gives us some information about how the company sees the future of this type of content. In June 2026, the company released a free AI music detector for playlists, available in 27 languages, that allows any person to scan her or his playlist on 20 different platforms, including Spotify and Apple Music, and discover the amount of artificial music in it. Deezer says that 43 percent of users who transfer from other platforms to Deezer come with the playlist that includes AI-generated music. This was, at least partially, a competitive move, but it is a very useful tool, and I ran my own playlists through it.

Here is one more Deezer-related information: the company started selling its detection technology in January 2026, and the first client was France’s royalty collection society Sacem. When the fraud detection arm of a streaming platform starts being sold as a separate product, you understand that the problem is not going anywhere.

What Is Written in TIDAL’s June 2026 AI Policy

It took longer for TIDAL to react, but, when the company decided to act, it decided to go beyond others in one particular aspect. TIDAL published its full AI policy on June 29, 2026 under the title Promoting Fairness and Economic Empowerment in the Era of AI-Generated Music, describing it as a living document that will be updated as the technology develops.

The key statement of the policy is very clear: the royalties should be earned only by the original works that were directly produced, written, and performed by people. As reported by Music Business Worldwide, the demonetization of wholly AI-generated music started on the day when the policy was published, while the introduction of the consumer-friendly AI badge and removal of fraud started on July 15, 2026. Since that date, all the tracks identified by the company’s systems as 100 percent AI-generated earn nothing. Not the reduced payment. Nothing.

TIDAL is not blocking uploading, streaming, and discovery of this type of tracks. The company says explicitly that artists should be free to use the AI tools for their creations, and listeners should be able to choose their own music freely. AI tracks can be uploaded, streamed and discovered, but they cannot be monetized, and they cannot be sold through direct fan sales. The same rules apply to independent artists through TIDAL Upload, direct uploading program of the platform, so, there is no side door.

In addition to these, here are three more points of the policy that matter in practice. First, the tagging: now listeners see an AI badge next to the content that was identified as fully AI-generated, and the company promises to expand this badge on substantially AI-generated content as the detection methods become more accurate. Second, the enforcement of fraud: the AI tracks that impersonate real artists, exploit somebody’s name or likeness, deceive listeners, and have suspicious uploading and streaming patterns are blocked or removed, not just demonetized. Third, and this is the hidden point, TIDAL states that it expects and will begin to enforce identification of AI-generated content by content distributors before it reaches the platform. This will reconfigure how distributors screen uploads for any platform, not only for TIDAL.

Tony Gervino, TIDAL executive, explained the motivation in a very simple way: many listeners told the company that they did not want to be exposed to, or recommended, wholly AI-generated music. As was reported by DJ Mag, the policy also leaves a door open, saying that TIDAL will reconsider how it treats monetization of AI-assisted tracks built with the help of properly licensed models as industry standards evolve. This provision matters more than it might seem, as the issue of training data licensing is still not resolved.

Real Differences Between the Policies of the Two Platforms

People keep comparing Deezer and TIDAL’s approaches and calling them the same. But they are not, and if you create music, then you should know the differences and what do they mean for your royalties.

TIDAL’s rule is categorical: if the track is identified as 100 percent AI-generated, it earns zero royalties regardless of whether the streams are humans or bots. The fully AI track with the huge number of fans will earn nothing on TIDAL. This is a philosophical statement about what deserves payment, not the fraud measure.

Deezer’s rule is surgical: it demonetizes the fraudulent streams and hides the AI tracks in the discovery system, but the legitimate play of an AI track can still receive royalty according to the current version of the policy. In January 2026, Deezer focused on demonetizing the fraud, that was 85 percent of AI streams, in its public statements. The practical outcome is similar as the number of real listeners who stream this type of music is so small, but the principle is different.

The other difference is the visibility of the intention. Deezer developed and patented its own detection technology, sells it commercially, and publishes quarterly upload statistics like a weather forecast. TIDAL has not said anything about how its detection actually works, which is why Bill Rosenblatt’s analysis in Forbes called the policy of the platform one of the most stringent of any music service, while noting that the burden falls heavily on detection technology that was not yet fully proven at scale.

Comparison with Spotify and Apple Music, for Context

Contrast with the two largest platforms makes the differences between TIDAL and Deezer clearer. While Spotify still pays recording royalties on AI music in the usual way, based on stream economics, its September 2025 policy was focused on unauthorized voice clones, spam filter for behavioral abuse, and adoption of the DDEX disclosure standard, which is an industry metadata format, that allows artists to declare AI usage in their credits. Spotify said that it removed more than 75 million spammy tracks in the year before that announcement, but the responsible usage of AI in music is still payable there. Apple Music introduced the AI tags in March 2026 based on the identification made by labels and distributors and pays AI tracks normally, while aggressively fighting the royalties associated with fraud. So, as of July 2026, the same track can earn full royalties on Spotify and Apple Music, earn nothing on TIDAL, and be labeled, hidden and partly demonetized on Deezer. TechTimes pointed this out exactly when RIAA and a coalition of industry organizations suggested voluntary AI-Generated and AI-Assisted labels in the week when TIDAL demonetization became effective: the rules differ so widely that earning potential of the track now depends only on the platform that judges it, without changing the music.

What Does It Mean if You Use AI Tools in Your Music

First, the good news. Both platforms draw the line at wholly or fully AI-generated content. If you write, perform and produce your music and use AI as the tool for that, whether it is stem separation, mastering, artwork creation, or generating textures that you play over, you can earn royalties on all platforms. TIDAL’s policy targets what I would call prompt and publish content, where a human types a description and uploads whatever comes out. Human-led work with AI assistance is explicitly allowed and paid.

Second, the honest warning. There is no definite boundary between assisted and generated in TIDAL’s policy, and the detection systems sometimes make mistakes. If your workflow heavily uses generative tools like Suno and Udio for the actual composition and performance, you should assume that Deezer’s detector will identify it as fully AI-generated, because Deezer specifically trained its system on the output of these models, and you should assume that TIDAL will demonetize it. Plan your revenues correspondingly and consider these two platforms promotion platforms rather than income ones.

Third, check your distributor before you check any platform. Forbes reported that TuneCore now only accepts fully AI-generated music made with platforms that used fully licensed datasets, while CD Baby bans fully AI-generated music outright and terminates accounts that repeatedly try to upload it. Your music will be rejected before Deezer or TIDAL ever sees it. One word of warning from somebody who reads terms of services for a living: there are services that openly sell tools to strip AI detection artifacts from tracks so that they pass platform checks. Do not even think about them. Disguising AI content to get the royalties that were explicitly banned by the platform is exactly the kind of activity that fraud policies were introduced to punish, and account termination is one of the mild penalties.

Fourth, do the math on what you are losing. According to Royalty Exchange’s 2026 platform comparison, TIDAL pays roughly 0.013 to 0.015 dollars per stream, the highest rates in streaming industry, while Deezer is around 0.005 to 0.007 and Spotify is 0.003 to 0.005. TIDAL’s per-stream rate is higher, but the audience of the platform is a fraction of Spotify’s. The practical loss of income from the policies of these platforms is smaller than you can expect from the headlines, because the bulk of the streams and money still goes through the platforms that pay normally. The real danger is in the trendline, because TIDAL just proved that a major service can demonetize AI content without collapsing.

What Does It Mean for Human Artists

If you create music in the old-fashioned way, both policies work in your benefit, and Deezer’s numbers prove that the protection is real rather than symbolic. With the AI tracks being limited to 1 to 3 percent of the streams and the fraudulent majority of them being demonetized, your dilution of royalties on Deezer is minimal. On TIDAL, every dollar that would go to the prompt-generated content now stays in the pool, and TIDAL’s above average per-stream rate makes this pool worthy of defending.

The research confirms the direction too. Deezer commissioned Ipsos to conduct a survey of 9,000 people in eight countries at the end of 2025, and the results were sobering. Ninety-seven percent of participants could not distinguish AI-generated songs from human songs during a blind test. Eighty percent thought AI songs should be clearly marked, 73 percent of streaming users needed information about recommendations of synthetic songs in their streaming service, and 52 percent found that completely AI-generated songs should not go into the same charts as human ones. The listeners cannot notice any differences, and they understand that. This is why they want the streaming platforms to do that for them.

It seems to me that what human artists should do is unexciting – they should check whether their work gets distributed through Deezer and TIDAL. Independent artists often omit these two platforms because the audience is smaller than it is on the others and start complaining about how AI fills Spotify. Now, these two streaming services do their best to secure your share of the market pool and pay one of the best prices for your music.

The Problem That Is Not Going Anywhere by Declaration

This is what I see as an honest reality check because every statement in this article depends on one key point: can the platforms reliably recognize AI-generated songs? Deezer claims to have patents and train its detection methods using the main generative models. But this company sees the task as finding the most popular tools, and new models require new training data. TIDAL did not publish anything about their detection stack but promised to expand tagging to include substantially AI-generated content as the methods improve. This means that they are not there yet.

False positives are the most terrifying nightmare scenario. Human artist who gets their track recognized as 100 percent AI on TIDAL loses all their royalties on the platform. And as for now, the policy of the company does not mention any appeal procedures in case the song was classified incorrectly. False negatives are the opposite case – fraud operators are flexible and quick; Michael Smith managed to run his operation for seven years on four major platforms before getting caught. Platforms bet that imperfect enforcement is better than no enforcement, and I think they are right, but if you are an artist reading this article, please keep copies of your sessions, stems, and project timestamps. Proof of human authorship is about to become a real practical advantage.

What I Would Actually Do This Week

If you are an artist, run the Deezer playlist detector on your profile playlists and the biggest editorial playlists of your genre just to understand the situation you are playing in. Make sure that your distributor distributes your song through Deezer and TIDAL. If you use AI somewhere in your production process, document that and find out if your distributor supports DDEX AI disclosure fields – it will save you money and energy in the future.

If you create songs heavily using AI technologies, analyze your business model. The era of anonymous uploading earning passive royalties on the platforms is closing, one by one. Direct licensing, sync placement, and transparent AI projects still have place. Not pooling farming.

If you are a user, the tags are already active. Deezer marks AI-generated content on the track level, TIDAL introduced the icon in mid-July, and for the rest of the platforms, there is a free AI detector working on 20 platforms. It is the first time in the history of generative music when we are able to actually see what we are listening to. It remains only to wait and see whether other platforms follow Deezer and TIDAL with their royalties.

Sources

Common questions

How do Deezer and TIDAL treat AI music royalties differently?

TIDAL cuts royalties at the track level, so any track its systems identify as 100 percent AI-generated earns nothing regardless of who streams it. Deezer acts at the stream level, tagging AI tracks, excluding them from recommendations and editorial playlists, and demonetizing the fraudulent streams, which is about 85 percent of AI streams.

Does TIDAL pay royalties on AI-generated music?

No, since its policy published June 29, 2026, TIDAL will never knowingly pay royalties on wholly machine-generated tracks, and demonetization began that day with an AI badge and fraud removal following on July 15. AI tracks can still be uploaded, streamed, and discovered, but they cannot be monetized or sold through direct fan sales.

Does Deezer demonetize all AI music?

No, Deezer does not demonetize every AI track automatically, and a listener who searches for and streams an AI track can still generate a royalty under the current policy. What Deezer removes from the pool is the fraud, and since fully AI-generated music is only 1 to 3 percent of streams with up to 85 percent of those demonetized, the impact on the pool is minimal.

Why are streaming platforms cracking down on AI music royalties?

The driver is fraud and the sheer scale of uploads, with Deezer receiving about 90,000 fully AI-generated tracks daily, now more than half of all uploads. Deezer found up to 85 percent of streams on fully AI tracks in 2025 were fraudulent, and in March 2026 one man pled guilty to earning more than 10 million dollars in royalties using AI tracks and streaming bots.

Can I still earn royalties if I use AI tools in my music?

Yes, both platforms draw the line at wholly AI-generated content, so if you write, perform, and produce your music using AI for tasks like stem separation, mastering, artwork, or textures, you can earn royalties everywhere. TIDAL's policy targets prompt-and-publish content, while human-led work with AI assistance is explicitly allowed and paid.

Will Deezer and TIDAL detect music I made with Suno or Udio?

If your workflow heavily uses generative tools like Suno and Udio for the actual composition and performance, you should assume Deezer's detector will flag it as fully AI-generated, since Deezer trained its system on those models, and that TIDAL will demonetize it. The article suggests treating these two platforms as promotion rather than income in that case.

How do Spotify and Apple Music handle AI music compared to Deezer and TIDAL?

Spotify still pays recording royalties on AI music normally, focusing instead on voice clones, spam, and DDEX disclosure, and Apple Music also pays AI tracks normally while fighting fraud. So the same track can earn full royalties on Spotify and Apple Music, earn nothing on TIDAL, and be labeled, hidden, and partly demonetized on Deezer.

How much do streaming platforms pay per stream in 2026?

According to the comparison cited, TIDAL pays roughly 0.013 to 0.015 dollars per stream, the highest in the industry, while Deezer is around 0.005 to 0.007 and Spotify is 0.003 to 0.005. TIDAL's rate is higher, but its audience is a fraction of Spotify's, so the practical income loss from these AI policies is smaller than headlines suggest.

What is the biggest risk with AI music detection systems?

The most serious risk is false positives, where a human artist's track is wrongly classified as 100 percent AI on TIDAL and loses all royalties, with no appeal procedure currently mentioned in the policy. Because of this, the article advises keeping copies of your sessions, stems, and project timestamps, since proof of human authorship is becoming a real practical advantage.