All articles

Will AI Music Be Banned?

I release music. Some of it is recorded on real instruments in a real room, and some of it utilizes AI software for stems, sketches, and sometimes even complete demos. And so when someone asks me whether AI music will be banned, they’re not asking an abstract policy question. They’re asking whether the songs I made last weekend are about to disappear from Spotify, or whether the tsunami of machine-generated lo-fi music drowning my playlist is about to dry up.

Here’s the straightforward answer upfront. No, AI music is not going to be banned outright, not in the United States, not in the European Union, and not on the major streaming platforms. Here’s what is actually happening instead: individual platforms are setting their own policies, from a full ban on Bandcamp to demonetization on Tidal to tagging on Deezer. Courts in Boston, New York, and Munich are hearing cases on whether the companies behind AI music generators infringed copyright laws when they built them. And lawmakers are proposing disclosure regulations, not prohibition of AI music. If you release AI music, your world is changing beneath you. If you hate AI music, there is relief coming, but not of the sort you probably imagine.

Follow me through the details below, which I’ll provide you backed up with numbers and dates, because there is almost no serious commentary out there on this issue.

What Exactly “Will AI Music Be Banned?” Means

If you type “will AI music be banned” into the search engine, most of the time you’re asking one of three different questions, and the answer differs in each case.

First, are governments going to prohibit AI-generated music. This is not going to happen, neither in the United States, nor in the European Union, and I know of absolutely no such legislation proposed anywhere. Second, are the streaming platforms going to remove AI-generated music. Some of them already did, selectively, and at least one of them prohibited it entirely. Third, are the tools themselves – Suno, Udio and their competitors – going to be shut down through lawsuits. In 2024 it was a real possibility. Today it is not, because of the record labels decided to license them instead.

Remember this three-thread distinction as you keep reading, because confusing them is the easiest way to believe bullshit.

The Flood That Prompted This Question

That is what prompted this question, and it explains almost every single step taken since. In order to see the flood, you need to look at the numbers from the streaming platforms.

Deezer is currently the only major streaming platform to operate a publicly available AI detector, and the statistics it published are telling, to say the least. In January 2025, Deezer saw about 10,000 fully AI-generated tracks daily. By April 2026 the number grew to 75,000 daily, which is 44 percent of all daily uploads, or more than two million AI music tracks per month. And in July 2026 Deezer reported that for the first time more than 50 percent of daily uploads to the platform were AI-generated, peaking at around 90,000 tracks per day in June. Let me say it once more. At peak days, more than half of all “music” being uploaded to a major streaming platform is machine-generated.

But the most important part of this news came later, because this huge number of uploads is deceiving. According to Deezer, fully AI-generated tracks account for only 1 to 3 percent of all streams on the platform. And up to 85 percent of those streams were determined to be fraudulent in 2025, in the sense that the music was uploaded by bots to the bots who play it. And this is the real story here. Most of AI-generated music is being uploaded to streaming platforms not because someone wants to listen to it. It is uploaded as part of a royalty fraud scheme.

Spotify’s numbers paint a similar picture. According to Music Business Worldwide, Spotify deleted more than 75 million spammy tracks in the twelve months leading to the September 2025 announcement of its new policy regarding AI music. Seventy-five million! As context, Spotify’s music catalog wasn’t even close to 100 million tracks not long ago.

Next time you hear about “AI music crackdown,” make sure to remember what it really is. Industrial-scale spamming and fraud, not your neighbor’s Suno experiments.

Platforms’ Positions Right Now

This is where you’ll find a concrete answer to your question, because decisions on the platform side have already been made, and they’re scattered widely.

Bandcamp was the most aggressive. On January 13, 2026, in its policy “Keeping Bandcamp Human” the indie music marketplace announced that music generated wholly or in substantial part by AI is no longer permitted, as well as any use of AI to impersonate other artists or music style. As Ars Technica reported, Bandcamp is the first major platform to completely ban AI-generated music. If your workflow is prompt-to-song, Bandcamp is out of the question. The “substantial part” formulation is important too. It allows Bandcamp to delete tracks where AI did most of the work even if it was then edited by humans.

Tidal chose economics instead of removals. At the end of June 2026, Variety reported that Tidal is going to require artists to mark the music they created wholly or substantially using AI, and such tracks would not be eligible for royalties, starting July 15, 2026. Tidal is also prohibiting AI-generated music linked to fraud or impersonation. You can release AI-generated music on Tidal. You can just not get paid for it.

Deezer tags and isolates. AI-detecting tracks remain on the platform but are excluded from algorithmic recommendations and editorial playlists. And from June 2025, the listeners can see an AI tag on such tracks. In July 2026 Deezer adds some enforcement: AI music involved in streaming fraud will be deleted from the platform, and so will be AI-generated tracks which don’t receive any plays in six months. This latter rule is especially tough for content farms because most spam tracks receive almost no organic plays.

Spotify, the platform everyone cares about, explicitly does not ban AI-generated music. The framework, introduced by Spotify in September 2025, consists of three components: a stronger impersonation policy that prohibits unauthorized AI voice clones, a spam filter aimed to tackle mass uploads and modified metadata, and the support of new AI disclosure standard developed through DDEX, the organization which sets music metadata standards. According to Spotify executives, an AI act like The Velvet Sundown, which fooled hundreds of thousands of listeners in 2025, will be allowed on Spotify, though encouraged to label itself. And Apple Music has an optional AI tagging system and remains relatively quiet.

There are also smaller human-first platforms. FriendsGroove, which launched in July 2026, forbids AI-generated tracks and voice cloning and has every artist approved manually by humans. You can expect to see more niche platforms positioning themselves in this manner in the future, because “certified human” is becoming a selling point.

See the pattern? None of these is a legal ban. All of them are simply platform’s policy on what it hosts and pays for. That’s how the future looks like: patchwork, not wall.

The Lawsuits That Might Have Shut Down the Tools

By far the most serious threat to AI music ever happened in the form of lawsuits filed by the RIAA, the trade body for the US major record labels, against Suno and Udio, the dominant AI-generated music tools, on June 24, 2024. The labels argued that both companies trained their models on copyrighted music tracks without their permission, and sought statutory damages up to 150,000 dollars per infringed recording. Multiply this by dozens of thousands of recordings and you get numbers that literally kill the companies.

But then something interesting happened. Starting in October 2025, the labels started settling with Udio and Suno and signing licensing agreements with the companies they sued. Universal Music Group settled with Udio in October 2025. Warner Music Group settled with both Udio and Suno in November 2025. As Virginie Berger explained in her Forbes article, this process can be described as launch, train, settle, license, where the labels turn infringement claims into licensing fees while independent artists, whose music the companies used to train their models, received nothing. Udio is now rebuilding as a licensed, closed garden service planned to launch in 2026. But Sony Music Group has no settlements yet and continues the lawsuit against both Udio and Suno. Universal’s case against Suno is also ongoing in Boston, where the labels moved in May 2026 expanding it from 560 infringed works to more than 61,000 after finding out in discovery that Suno used millions of their tracks to train.

Settling created another round of lawsuits, which shows how raw the situation is now. At the end of June 2026 the American Federation of Musicians, US musicians’ union, sued Universal and Warner for licensing recordings to Suno and Udio without sharing the licensing fees with the musicians who actually performed on them. Class action lawsuits are also filed by independent artists, because settlements covered only major labels’ catalogs.

Watch Munich for another verdict. The German collecting society GEMA sued Suno in January 2025 for compositions including Forever Young and Mambo No. 5, and the verdict by Munich Regional Court is scheduled for July 31, 2026, delayed from June. The same court previously ruled against OpenAI in November 2025 in the similar case on song lyrics. A GEMA victory will be the first major European ruling on AI platforms needing licenses for training on copyrighted music, and European courts tend to be friendly towards creators compared to US courts.

Here’s my take as someone who follows these lawsuits closer than needed. The lawsuits are not going to ban AI music. The lawsuits are going to decide who will get paid for it. Every single settlement is turning a potential ban into a licensing fee, and the lawsuits left are fights over the fee size and distribution. The tools survive in any case. What will change is their licensing price, training data and the catalogs they can imitate.

What Governments Are Really Doing

If you’re looking for the Congress or Brussels to make AI music illegal, you’re wasting your time. The legislative activity is headed in a completely different direction: identity protection and disclosure.

On the identity protection, the landmark piece of legislation is the Ensuring Likeness Voice and Image Security act, aka the ELVIS Act, passed in Tennessee and effective since July 1, 2024, which creates an enforceable right to personal names, pictures, voices and likenesses, passed unanimously and as a direct reaction to AI voice cloning. At the federal level, the NO FAKES Act would extend this right to all US citizens, allowing artists to control AI-generated replicas of their voice and likeness. The legislation was reintroduced and received broad industry support but still has not passed as of the middle of 2026. These pieces of legislation prohibit a specific abuse of AI technologies, not the technologies themselves.

On the disclosure, the Senate introduced the AI Labeling Act of 2026 on June 25, which would require AI-generated audio, video and image to carry both a visible disclosure and machine-readable metadata identifying the system that created them. In Europe, transparency requirements of Article 50 of the EU AI Act are going to become effective on August 10, 2026, requiring providers of AI generators to mark synthetic content as artificial and creating a code of conduct on marking and labeling, which is expected to be finalized through 2026. For music specifically, that means watermarking and metadata, not removal. And there is a loophole worth remembering: when the work is evidently artistic or creative, the disclosure may happen in a way not spoiling enjoyment of the work, so nobody will start the track with a robot voice introducing itself.

Another legislative effort is focused on the money part. The Protect Working Musicians Act, reintroduced in May 2026, would allow independent musicians to negotiate collectively with streaming and AI platforms, effectively creating an antitrust exemption so musicians could bargain like major labels. Once again, this is about money, not ban.

One legal fact that surprises many users of AI music generation tools, and which I always mention every time someone asks me about releasing Suno music, is that, according to the existing Copyright Office guidance, a pure AI-generated song without any meaningful human contribution cannot be copyrighted and cannot be registered.

What The Music Industry Is Really Trying To Achieve

The clear signal of the direction of the developments came on July 10, 2026, when a coalition of eight organizations, including the RIAA, IFPI, Recording Academy, SAG-AFTRA, A2IM, WIN, IMPALA and the Human Artistry Campaign, introduced a voluntary unified labeling program for AI in music. The system introduces two tags, which are very loosely inspired by the familiar explicit content tag. “AI-Generated” tag goes to the tracks generated purely from a prompt and created by AI, or when AI generates lead vocals or primary instrumentation. “AI-Assisted” tag goes to substantially human-generated tracks where AI created some elements.

Look closely who is in this coalition. These are the organizations with the most motivation and with the strongest power to push for a ban, and what they asked for instead is a sticker. The Wall Street Journal reports that the groups plan to cooperate with Spotify and Apple Music in implementing the labels, and the critics of AI in the pro-AI camp rightly point out that this system conspicuously overlooks AI in composition and lyrics generation, the areas which major label artists already use the tools for. The industry is not trying to prohibit AI music. The industry is trying to fence off the market niches where human authenticity is a premium feature, while licensing the rest.

Whether the listeners will care about it is an open question, but the early evidence says that they do. Deezer conducted a survey on attitudes towards AI-generated music in the late 2025 and decided to tag AI tracks precisely because of that. I also personally noticed it when posting my AI-assisted creations. Nobody complains about my AI string arrangement. People complain about it when they feel deceived.

What This Means If You Generate Music With AI

If you use Suno, Udio, or other AI music generators and plan to release their music, here is the practical playbook for a friend of mine, based on the current state of rules.

Always disclose your usage from the start. Tag your tracks with AI usage through your distributor using DDEX metadata, and use the AI-Generated or AI-Assisted tags as platforms implement them. Every policy trend of the last eighteen months punishes hiding, not the usage. The artists being banned by the platforms are the ones pretending to be human-made.

Do not clone voices without written consent. This is the one true ban that already exists pretty much everywhere. Spotify removes voice clones on complaint, Tennessee bans it by law, and the NO FAKES Act would make it federalized. There is absolutely no gray area to explore here.

Do not flood the platform with music tracks. Deezer is deleting tracks which do not receive any plays for six months, Spotify’s spam filter hides mass uploaders, and the whole purpose of the AI-generated upload is to commit a fraud. Michael Smith was sentenced in July 2026 in a federal court in New York for using AI tracks and bots to generate roughly 10 million dollars in royalties in what is, according to prosecutors, the first criminal case of this kind.

Keep evidences of your human authorship. Keep a record of your prompts, your DAW session, your lyrics, your arrangement decisions. If copyright battles over AI music ever reach your catalog, the difference between “machine-generated content” and “human-authored work made with help of AI” will depend on your evidences. Write your own lyrics if you can, because a human lyric is independently copyrightable in addition to the instrumental.

And avoid Bandcamp for AI music altogether. Don’t even try to sneak around the policy, because the “substantial part” formulation is broad enough to catch you, and deletion from a human-friendly platform ruins your reputation more than the sale does.

What This Means If You Want AI Music to Go Away

If you are on the other side of this, fed up with the tsunami of algorithmically-generated tracks, the reality check I have for you is simple. The flood is not going away, because the tools are cheap, global and increasingly licensed. What you will get instead is the chance to see it and work around it.

Deezer already tags AI tracks and excludes them from algorithmic recommendations, and introduced a tool in 2026 which can detect AI-generated tracks in Apple Music and Spotify playlists. Bandcamp is human-only marketplace now by policy. Tidal deprived the flood of any economic benefit by cutting out AI-generated tracks from royalties. If Spotify and Apple implement the labeling standard, you will be able to filter by origin as you now filter explicit content. The economic pressure is more powerful than any ban could have been. Fraudulent AI-generated tracks are uploaded to the platforms in order to skim the royalties, and every demonetization makes the scheme less profitable.

Where I See All This Going

If you ask me whether AI music will be banned, my answer is “no.” But if you ask me whether it will be everywhere, unlabeled and unregulated, my answer is also “no.” The next two years will be about separation, not about ban.

The events to watch in the near term are specific. The Munich court is scheduled to deliver the ruling on GEMA v. Suno on July 31, 2026, and GEMA victory will mean licensing across Europe. The US fair use question in the remaining lawsuits by Sony and Universal won’t be decided earlier than 2027, and dispositive motions in the Boston Suno case won’t be delivered earlier than April 2027, according to the court dockets monitored by Chartlex. The transparency rules in the EU AI Act are going to bite in August 2026. And the big two streaming platforms are going to announce their policy decisions in the next few months.

My prediction, for whatever value a practitioner’s prediction could have: by 2028, the AI-generated music will be in its own labeled and largely demonetized lane on the big two streaming platforms, the licensed AI tools will be a normal part of every producer’s toolkit, the voice cloning without consent will be a legal violation in the majority of the developed world, and a few human-only platforms will thrive as farmers’ markets for music. This will not be a ban. This will be pricing structure.

If you make only one thing this week, make it this one. If you release music in any form, log in to your distributor and figure out how to label your tracks as AI-generated, if it is possible, and fill the information on the tracks that you’ve already released.

Sources

Common questions

Is AI generated music illegal?

No. There is no legislation in the United States, the European Union or anywhere else the article is aware of that would make AI music illegal. The laws being passed target specific abuses, such as Tennessee's ELVIS Act against voice cloning, and the proposed AI Labeling Act and EU AI Act transparency rules that require disclosure and metadata, not removal. The one legal wrinkle is that a purely AI generated song cannot be copyrighted under current US Copyright Office guidance.

Which streaming platforms ban AI music?

Bandcamp is the only major platform with a full ban, under its January 2026 Keeping Bandcamp Human policy covering music generated wholly or in substantial part by AI. Tidal allows AI music but pays no royalties on it from July 15, 2026. Deezer keeps it on the platform but tags it, excludes it from recommendations, and deletes AI tracks tied to fraud or unplayed for six months. Spotify explicitly does not ban it. Smaller human only platforms like FriendsGroove are emerging too.

Does Spotify allow AI music?

Yes. Spotify's September 2025 framework has three parts: a stronger impersonation policy against unauthorized AI voice clones, a spam filter for mass uploads and manipulated metadata, and support for a DDEX standard so AI involvement can be disclosed in credits. Spotify executives have said an act like The Velvet Sundown would be allowed, though encouraged to label itself. Spotify deleted more than 75 million spammy tracks in the year before that announcement.

How much music on streaming services is AI generated?

On Deezer, the only major platform with a public AI detector, fully AI generated uploads grew from about 10,000 a day in January 2025 to 75,000 a day by April 2026, and passed 50 percent of daily uploads in July 2026 with a peak near 90,000 a day. Yet those tracks account for only 1 to 3 percent of streams, and up to 85 percent of those streams were found to be fraudulent, meaning most AI uploads exist to skim royalties rather than to be heard.

Will the lawsuits shut down Suno and Udio?

Almost certainly not anymore. The RIAA sued both in June 2024, but starting in October 2025 the labels began settling and licensing instead, with Universal settling with Udio and Warner settling with both companies. Sony's cases and Universal's case against Suno continue, and the Boston case expanded to more than 61,000 works in May 2026. The article's view is that the lawsuits will decide who gets paid for AI music, not whether it exists.

What is the GEMA v. Suno case?

GEMA, the German collecting society, sued Suno in January 2025 over compositions including Forever Young and Mambo No. 5, with a Munich Regional Court verdict scheduled for July 31, 2026. The same court ruled against OpenAI in November 2025 in a similar case about song lyrics. A GEMA win would be the first major European ruling that AI platforms need licences to train on copyrighted music, and would likely push licensing across Europe.

Will AI music have to be labelled?

Increasingly, yes. The EU AI Act's Article 50 transparency rules take effect August 10, 2026, requiring synthetic content to be marked, though in a way that need not spoil artistic works. The US Senate introduced the AI Labeling Act in June 2026. And on July 10, 2026 a coalition including the RIAA, IFPI, Recording Academy and SAG-AFTRA proposed voluntary AI Generated and AI Assisted tags for streaming platforms, loosely modelled on the explicit content tag.

Can I release AI music on Spotify without problems?

Yes, if you follow the direction every policy has taken: disclose AI use through your distributor's DDEX metadata, never clone a voice without written consent, do not flood the platform with uploads, and keep evidence of your human authorship such as prompts, DAW sessions and your own lyrics. The artists being removed are those hiding AI use, cloning voices or running upload farms, not those using the tools openly.

Is streaming fraud with AI music a crime?

It can be. In July 2026 Michael Smith was sentenced in a federal court in New York for using AI tracks and bots to generate roughly 10 million dollars in royalties, which prosecutors described as the first criminal case of its kind. Platforms are also cutting the economics from the other side, with Tidal removing royalties from fully AI tracks and Deezer deleting unplayed AI uploads.

What will happen to AI music in the next few years?

The article predicts separation rather than a ban. By 2028 it expects AI generated music to sit in its own labelled and largely demonetized lane on the two biggest streaming platforms, licensed AI tools to be a normal part of producers' toolkits, voice cloning without consent to be illegal across most of the developed world, and a handful of human only platforms to thrive as a premium niche. Key near term events are the Munich verdict, the EU rules in August 2026, and pending policy decisions from Spotify and Apple.