The Verge: AI music floods streams
- Deezer says nearly 75,000 fully AI-made tracks now hit its service every day, or 44% of new uploads, as The Verge maps the spillover. - The weird part is demand is not following supply: Deezer says AI music gets just 1% to 3% of streams, and 85% look fraudulent. - That turns discovery into a spam problem — not a taste revolution — while labels and platforms scramble to tag, filter, and remove fakes.
Music streaming has a catalog problem now, not just a copyright problem. AI song generators have made it cheap and fast to upload endless tracks, so services are filling with synthetic music faster than listeners are asking for it. That gap is the whole story. The news this week is that The Verge pulled the pieces together just as Deezer’s latest numbers showed the flood getting much worse — nearly 75,000 fully AI-made tracks a day, or 44% of new uploads on its platform. ### Why is this suddenly everywhere? The tools got simple. A few years ago, AI music still looked like an experiment for hobbyists and producers. Now services like Suno and Udio can turn a text prompt into a finished song in minutes, which means the bottleneck is gone. If making music used to require skill, does. ### Are people actually listening to this stuff? Not much, turns out. Deezer says fully AI-generated tracks make up only 1% to 3% of total streams on its platform, even while they account for 44% of fresh uploads. That is the key mismatch. Supply is surging, but listener demand is not. It is less like a new genre taking off and more like junk mail stuffing the mailbox. ### So why upload so much of it? Because streaming rewards volume, and bad actors think they can game the system. Deezer says 85% of streams on fully AI-generated music are fraudulent and demonetized, which points to bots, fake engagement, and low-cost spam economics. If one human-made album takes months enough noise to catch weak spots in recommendation systems and royalty pipes. ### What does this do to real artists? It crowds the shelf and muddies search. That matters most for smaller musicians, niche scenes, and fans who rely on playlists or algorithmic discovery. TIME highlighted cases where AI tracks impersonated or piggybacked on real artists, and Sony Music said in March it music exists now.” It is also reputational harm, release-campaign sabotage, and diluted attention. ### Can listeners even tell the difference? Usually not. Deezer says 97% of people in its Ipsos-backed study could not reliably distinguish fully AI-generated songs from human-made ones, and 80% said fully AI-generated music should be clearly labeled. That is why labeling matters so much. If listeners cannot hear the boundary, platforms have to show it. Otherwise discovery becomes guesswork. ### Are platforms banning AI music? Not really. They are edging toward containment. Deezer tags fully AI-generated tracks and says it removes them from recommendations. Spotify announced in September 2025 that it would adopt AI labeling in credits, add a music spam filter, and ban unauthorized voice clone a moderation problem. ### Do listeners even want more AI music? The mood seems to be cooling. A recent Luminate study found U.S. consumer interest in listening to AI-assisted music fell from net -13% in May 2025 to net -20% in November 2025, with the sharpest drop among Gen Z and Gen Alpha. That makes the flood look even stranger: more AI music is arriving just as listener enthusiasm is slipping. ### What’s the bottom line? This is not really a story about robots replacing pop stars tomorrow. It is a story about distribution systems getting clogged by cheap synthetic supply. The hard problem for streaming services is no longer “can AI make a song?” It clearly can. The hard problem is whether platforms can keep search, playlists, and payouts usable when the cost of flooding the zone has basically collapsed.
