Companies are getting serious about flushing out musical AI slop

Earlier this month, Deezer, the Paris-based streamer reported that about 28% of all new music being uploaded to the platform is AI-generated. Deezer knows this because it worked with a software company that developed a detection program that’s pretty accurate. This week, Spotify removed 75 million songs as it goes after AI crap and impersonations.

Now Universal Music and Sony are teaming up with a research lab created at Stanford. SoundPatrol has developed a “forensic AI model for audio-video fingerprinting” that it says will be able to “identify the influence of original human-created music in fully or partly AI-generated music content.” SoundPatrol was co-founded by Michael Ovitz, the former president of Disney.

AI-generated music is creating many problems. A report from Spain this week said this: “AI could reduce authors’ royalties in Spain by as much as 28% by 2028 – equal to €100 million in that year alone, and €160–180 million between 2025 and 2028.”

Meanwhile, AI keeps getting better. Suno, a company being sued for billions for allegedly using copyrighted music to train its models without permission, has a new and better version of its program. Meta has Vibes, a way to “discover, create, and share AI videos.” We have ReSing, which is an AI tool that’s supposed to be for professional artists. And then there was Neon, which was a very bad idea. It’s gone out of business after a data breach.

Things are changing fast. But I’ll say it again: I don’t need AI to create art. I need AI to do all the crap I don’t want to do to leave me more time for creating art.