The voice is warm. The production is clean. The lyrics are about longing for something just out of reach, “into the blue” as one song has it, and the whole thing arrives pre-packed, like the idea of a record rather than one.
You add it to a playlist. You do not think twice. You were not supposed to.
AI-generated music is now flooding streaming platforms at a scale that is reshaping how music gets discovered, who gets paid, and what listeners think they’re hearing. Suno, one of the main tools making it possible, has spent the past two years building a billion-dollar company on the back of music it never licensed.
How It Sounds
Suno tracks in the neo-soul register tend to hit the same place every time: a chorus that arrives exactly where expected, a voice that sits correctly in the mix without having to fight for it.
The production breathes. Nothing is wrong. Nothing costs anything either.
What is actually missing is harder to name than people suggest. It is not rawness. Plenty of pristine, carefully controlled records are extraordinary.
It is more that real recordings carry the trace of decisions made under pressure. A vocal take kept because the singer was tired and it sounds that way. A bass part that sits slightly behind the beat because the player was listening to the kit and responding to it.
Suno does not have those pressures. The output is what the prompt asked for, which is not the same thing as what the music needed.
The tell in AI-generated vocals, at least currently, is in the consistency. A human voice over four minutes shifts in ways the singer does not choose: the slight thinning on a note held too long, the place where the consonant arrives slightly early because the phrase has been sung forty times and habit has crept in.
AI vocals stay level. They deliver. They do not reveal anything about who produced them, because no one did.
The Authenticity Argument, Which Is Mostly a Bad One
The quickest criticism of AI music is that it lacks soul because no one lived it. This is mostly rubbish, and the people deploying it should know better.
Brill Building writers in the early 1960s were churning out teenage heartbreak songs from offices in Midtown Manhattan. Carole King was 17. Barry Mann was writing about feelings he was actively manufacturing for commercial sale.
The entire Northern Soul scene ran on American records made by professional songwriters who had no particular connection to the people who ended up dancing to them in Wigan at 3am.
Ghost-writing has been standard practice in hip-hop for decades, not as a scandal but as an industry norm. If authenticity of origin were the test, the canon would need significant revision.
Where the AI case is genuinely different is structural, not moral. A ghost-writer makes choices. A cover version is an argument with the original.
Take what Al Green did to “How Can You Mend a Broken Heart,” which the Bee Gees recorded first and Green’s version simply dismantled. There is interpretation happening, someone hearing the song and deciding what it actually is.
AI-generated lyrics are pattern completion at scale. They produce the statistical shape of longing. The bridge about finding yourself. The chorus that confirms the verse without complicating it.
Not dishonest. Just without the friction that comes from someone actually trying to say something and finding it difficult.
Whether that matters is a different conversation. The more pressing problem is not that AI music lacks depth. It is that it floods a system already struggling to surface anything.
What Suno Did and How It Justified It
Suno is a Cambridge, Massachusetts-based AI music generator. Type a prompt, receive a track. By late 2025, it had been used by nearly 100 million people and raised $250 million at a valuation of $2.45 billion.
The training data that made this possible included, by Suno’s own admission in legal filings, “tens of millions of recordings” it did not license.
The RIAA sued in 2024 on behalf of major labels. Suno admitted the unlicensed use nine times across those filings, then argued the outputs were sufficiently different from the inputs to qualify as fair use.
Germany’s GEMA filed its own suit in January 2025. Denmark’s Koda followed in November. A class action from independent artists launched in June 2025 is ongoing.
Warner Music settled in November 2025: download caps, licensed models, Suno acquiring the Songkick concert discovery platform from Warner in the same deal. Universal and Sony are still in court.
Paul Sinclair, Suno’s chief music officer and previously a two-decade veteran of Warner/Atlantic, posted on LinkedIn after Grammy week 2025 to say he had come away “more convinced than ever” that protecting artists and empowering new creators were not in conflict.
He compared what Suno had done to the rise of the iPod and iTunes. “The relationship is often messy,” he wrote.
The relationship, in this framing, is a relationship between equals who had a disagreement. What actually happened is that Suno trained on tens of millions of recordings without asking, built a two-billion-dollar company, then settled with the parties large enough to force a negotiation.
The independent artists whose work almost certainly made up the majority of those tens of millions of recordings were not part of that settlement. They are still suing.
If something bad happens with an output Suno generates for a user, the contract makes clear it is the user’s problem.
This is the thing Sinclair’s LinkedIn post about empathy and natural cycles was written to obscure.
Key Questions About AI Music, Answered
Is Suno legal?
Currently, yes, though that is being contested in multiple courts. Suno argues its outputs do not reproduce the recordings it trained on and therefore do not infringe copyright. The RIAA, Germany’s GEMA, Denmark’s Koda, and a class action representing independent artists all disagree. Warner Music settled with Suno in November 2025. Universal and Sony are still litigating. The US Copyright Office’s May 2025 report questioned whether AI training on copyrighted music qualifies as fair use, particularly when the outputs compete commercially with the originals.
Is AI music stealing?
Suno admitted in legal filings that its model trained on tens of millions of recordings it did not license. Whether that constitutes theft in a legal sense depends on how courts rule on fair use. What is not disputed is that the music was used without permission and without payment. Independent artists whose work formed the majority of that training data have received nothing.
Does Spotify allow AI music?
Yes. Spotify does not ban AI-generated music and has no mandatory labelling requirement. It has introduced a spam filter targeting mass uploads and fraudulent streams, and encourages voluntary AI disclosure via a credits standard. It removed 75 million spammy tracks in 2025 but has not committed to excluding fully AI-generated content from recommendations. Deezer, by contrast, labels AI tracks, excludes them from algorithms, and withholds royalties from them.
Can AI music be monetised on streaming platforms?
Yes, on most platforms including Spotify. Suno users on paid plans can download and commercially distribute their generated tracks, uploading them to streaming services subject to monthly caps. Those tracks are eligible for royalty payments through standard distribution channels. Deezer is currently the only major platform that explicitly withholds royalties from fully AI-generated content.
What Spotify Was Already Doing Before This
Before AI music became the headline problem, streaming had already developed its own version of it.
Spotify had been placing low-cost tracks under pseudonymous artist names into algorithmically curated playlists under the banner of “perfect fit content,” accumulating streams without paying out at label rates.
Liz Pelly documented this at length in her 2025 book Mood Machine, and the practice predates generative AI by years.
The royalty pool model, where payouts go proportionally to total streams rather than to the fans of specific artists, was already being gamed by content farms running short tracks timed to hit the 30-second monetisation threshold.
The point is not that AI music broke something clean. The point is that AI music arrived into a system already running a version of this logic, and made it faster and cheaper by an order of magnitude.
The Numbers That Are Actually Strange
Deezer has confirmed that around 28% of daily uploads to its platform are fully AI-generated. Those tracks account for 0.5% of actual streams.
Spotify, which removed 75 million “spammy” tracks in 2025 alone, has since announced a spam filter, voluntary AI disclosure credits, and tightened impersonation policies.
The measures came after AI songs appeared on the profiles of Here We Go Magic (inactive since 2015), Sophie (died 2021), and Blaze Foley (died 1989).
A fake AI act called The Velvet Sundown accumulated over a million monthly listeners before updating its bio to acknowledge it was “a synthetic music project guided by human creative direction.”
Sienna Rose, the neo-soul profile at the centre of the LA Times story that circulated earlier this year, had 45 songs uploaded in eight weeks, 4 million monthly Spotify listeners, and around 7,000 Instagram followers.
Sleep Token has 5.7 million monthly listeners and individual songs sitting between 150 and 288 million plays. The relationship between streaming numbers and social presence across a real artist’s career follows a logic that Sienna Rose does not fit.
Spotify’s spam filter will demote rather than remove. It will not mandate AI labelling, only encourage voluntary disclosure.
Deezer goes further: it labels AI-generated tracks, excludes them from editorial and algorithmic recommendations, and withholds royalty payments from them entirely. Spotify has not committed to any of this.
One of Spotify’s co-CEOs addressed the AI music wave directly, describing a growing catalogue as “always very good for us” and stating that AI-generated music “attracts new users, drives engagement, and builds fandom.”
The cultural moment, the argument went, always happens on Spotify regardless of where the music was made. Bigger catalogues, more users, more advertising revenue.
That is a business position, not an artistic one, and it explains the gap between what Deezer is doing and what Spotify is not.
What the Bio Still Says
The specific charge against Suno is not that it makes music. It is that it built a business worth billions on recordings it never paid for, structured its contracts to pass legal liability to users, settled only with parties large enough to force it, and is currently describing all of this as industry evolution.
There is a separate question sitting underneath the copyright cases, and it concerns Sienna Rose’s bio specifically.
Claiming that a non-existent person sings “every word with a sense of truth and beauty” is not just unlabelled AI content. It is an active misrepresentation to consumers.
Music attorney Crystal, representing independent artists in the ongoing class action against Suno and Udio, has described this as consumer fraud: lying to the public about what something is and where it came from.
The copyright question asks whether the training was legal. The consumer fraud question asks whether the product is being sold honestly. Those are different fights, and the second one may be harder for the platforms to dismiss.
Guitar educator and YouTube commentator Rick Beato, who has been tracking the AI music wave closely, put the transparency argument in the simplest terms: you want to know what you are eating, which is why ingredients go on packaging. Twizzlers says artificially flavoured. Sienna Rose’s Spotify bio does not.
Beato’s full breakdown of the Sienna Rose case is worth watching, as is the detailed examination of Suno’s legal position and what its chief music officer’s public statements actually mean in context.
The music sounds like music. It sounds like the idea of neo-soul. It sounds like an enormous amount of recorded longing processed and returned in the shape of longing.
The bio for Sienna Rose still describes her as “a storyteller at heart, singing every word with a sense of truth and beauty.” She has never been in a recording studio. She has never been anywhere at all.
You might also like:
- AI Music Settlements: Labels Partner With Suno & Udio
- AI Songs Top Charts: Breaking Rust and the Rise of AI Slop
- Breaking Rust AI Exposed Country Music’s Algorithm
- Streaming Payouts 2025: Which Platform Pays Artists the Most?
- Spotify Price Increase 2026: AI Music and Artist Survival
- How Much Do Artists Make on Spotify in 2025?
If you are an independent artist whose music may have been used to train an AI model, IndieAiLawsuit.com is currently accepting claims as part of the ongoing class action against Suno and Udio.