How Do Music Streaming Platforms Use And Train Their AI?

In 2024, SoundCloud rewrote its Terms of Use and added a clause that lets the platform feed uploads into artificial intelligence tools working inside its service, according to tech ethicist Ed Newton-Rex. The text says users “explicitly agree” that their tracks may “inform, train, develop or serve as input to” machine learning features that help SoundCloud run.

Many artists only learnt about the change when Newton-Rex posted the wording online, and frustration spread through musician forums. Artists argued they had no warning and feared losing control of their work once it entered machine-learning pipelines.

SoundCloud answered through statements to TechCrunch and Reddit. The company said it has never passed tracks to outside labs and that the update only clarifies how in-house recommendation or playlist tools depend on data science.

To calm fears, SoundCloud added a “no AI” tag that creators can attach to uploads, signalling that third parties must keep hands off, according to SoundCloud’s Reddit post. The firm also promised that later tools will serve discovery, rights protection and extra income for artists.

 

How Is SoundCloud Using AI Right Now?

 

SoundCloud already pours machine learning into everyday listening. Its engine lines up new tracks, builds playlists and detects fraudulent streams, the company told TechCrunch.

In 2024 the platform launched a suite of creative features. These features remix songs, craft backing beats and even sing synthetic vocal lines. SoundCloud presented them as optional kits that put experimentation in the hands of creators.

Users worried that those aids would learn directly from their own uploads. SoundCloud insisted that no outside scraping happens and that it has not opened catalogue data to external labs. According to SoundCloud, only in-house data steers any generative features it may deploy.
 

 
But the fine print says:

“In the absence of a separate agreement that states otherwise, You explicitly agree that your Content may be used to inform, train, develop or serve as input to artificial intelligence or machine intelligence technologies or services as part of and for providing the services.

“For the avoidance of doubt, neither SoundCloud nor any third party is allowed to [i] use, copy or reproduce any Content delivered to the Platform under separate agreements, which is owned or controlled by third party rights holders (including artwork, images, logos, audio and audiovisual recordings (and any part thereof), underlying musical works and lyrics, and metadata) for the purposes of informing, training developing (or as input to) artificial intelligence technologies without authorisation from the applicable rightsholders.”

 

What Actions Is Deezer Taking Against AI?

 

Later last year, Deezer became the first streaming platform to sign the Statement on AI Training, according to the company. The manifesto declares that unlicensed use of creative work for model training is “a serious, unjust threat to the livelihoods of the people behind those works.”

Chief executive Alexis Lanternier said artists and songwriters sit at the heart of the business and must be paid fairly. He added that Deezer will keep guarding their catalogues through new payment models, AI detection and anti-fraud research.

In the beginning of this year, Deezer announced a detection tool able to tag music built with generators such as Suno and Udio. Patent filings in December cover the technique, which found that about 10 000 fully synthetic tracks reach the service each day, roughly 10% of daily deliveries.

Once flagged, those tracks vanish from algorithmic and editorial suggestions so that users see clear labels and artists keep chart space. Deezer explained that the tool can adapt to new models without retraining on specific data sets, giving it a larger guard against copycats.

A team of 15 specialists at Deezer has spent a decade tuning machine learning that spots stream manipulation. The company says the same research now feeds the artist-centric payment scheme launched in 2023, which channels royalties toward performers who attract genuine fans.

 

Where Do These Conversations Need To Start Leading To?

 

The 2 platforms are on way different paths, in terms of artificial intelligence in music. SoundCloud tells creators that in-house algorithms help listeners find songs while giving them the chance to block external training. Deezer, on the other hand, campaigns against any unlicensed model building and wants clear tags for synthetic tracks.

Both methods have the same goal, and that is guarding human work while keeping tech perks that listeners enjoy. Recommendation engines, anti-fraud checks and creative kits work using data science, and the labels and writers behind each song want guardrails against misuse.

More platforms could copy Deezer’s public stance if supporters judge the Statement on AI Training as an easy test of loyalty to creators. At the same time, SoundCloud’s promise of later opt-out tags shows a path for self-service control inside a single app.