This morning, you put on a chill playlist and a voice catches your ear. The problem is: behind some tracks that are already playing millions of times, there isn't a songwriter in a studio, but a machine churning out verses and choruses on demand. Songs attributed to almost phantom names, like Scarlyy2 with "Banc de touche" or Aventhis with "Mercy on My Grave," have slipped into the recommendations and found their audience, often without being identified as artificial productions.
Everything is happening very fast. With tools like Suno or Udio, a few words of instruction are enough to generate a complete song in seconds, vocals and arrangements included, with an increasingly convincing sound. The first attempts "sounded off," repetitive, compressed, and easy to spot. Today, the line is blurring: quality is rising, detection is becoming a puzzle, and streaming, the number one way to consume music, is a major accelerator for this mass-produced music.
Spotify lets it happen, Deezer shows its hand.
Faced with this wave, the platforms are eyeing each other warily. Spotify, the world leader, refuses to make AI an enemy in itself and prefers to target abuses: "The use of AI in itself is not a problem. What we penalize are abuses such as identity theft, unauthorized cloning, and fraud," explains Romain Takeo Bouyer, head of content analytics at Spotify. The problem for the listener can be summed up in one sentence: nothing currently obliges the major platforms to clearly indicate that a track is AI-generated, and this gray area benefits those who flood the catalogs.
Deezer takes the opposite approach, promoting the labeling of "100% generated" content as a choice of transparency. Behind this clash lies a very real concern within the industry: what have the models been trained on, what catalogs, what permissions, what compensation, and to what extent can a voice be imitated without permission? With so many tracks produced at low cost and in large volumes, the temptation for fraud also looms, because in streaming, every listen translates into disputed pennies.
In France, Sacem is pushing for a framework based on traceability and remuneration, while European discussions surrounding the AI Act are fueling demands for stricter rules on the transparency of content used and distributed. The debate goes beyond technology: it touches on the moral contract between platforms, creators, and the public—the contract that ensures we know who sings, who writes, and who receives the money. One prospect is already emerging: tomorrow, trust could become as important an argument as sound quality, and platforms will have to choose sides.
Community
Comments
Comments are open, but protected against spam. Initial posts and comments containing links undergo manual review.
Be the first to comment on this article.