Spotify quietly removed 75 million tracks last year. Seventy-five million. And that number is still growing, because tens of thousands of AI-generated songs are being uploaded to streaming platforms every single day. Now the music industry’s biggest players have decided enough is enough.
A coalition including Sony Music, Universal Music Group, and Warner Music Group has called for AI-generated tracks to be disqualified from global music charts. The group, which also includes smaller labels and independents, is asking chart companies like Billboard and Spotify to adopt stricter rules requiring that charting songs be primarily human-made. Any AI involvement, they argue, must also be demonstrably lawful, which is a pointed reference to ongoing litigation against tools like Suno and Udio, both of which face lawsuits from major labels alleging unauthorized training on copyrighted material.
This isn’t just about artistic integrity. Synthetic tracks gaming streaming algorithms represent a real financial threat to labels and human artists, because chart placement drives playlist placement, which drives royalties. If AI-generated songs can flood the top of Spotify’s most-listened lists or Billboard rankings without restriction, the incentive structure for investing in human artists starts to break down. That’s the core economic argument here, and it’s a serious one.
The push fits into a broader industry effort to force transparency onto AI music. Earlier this month, the RIAA and IFPI proposed a labeling system that would flag songs as fully or partly AI-generated directly inside streaming apps. Apple has already introduced AI “Transparency Tags” that work similarly, though applying them is currently optional for labels and distributors. That voluntary approach is exactly the kind of half-measure the coalition seems to be reacting against.
But detection is the hard part. There’s no reliable, universal method for identifying AI-generated audio at scale. Chart companies would need auditing infrastructure they don’t currently have, and AI music tools are improving fast. Google’s Lyria Pro 3 now generates tracks up to three minutes long, with distinct song sections like intros, choruses, and bridges. The outputs are getting harder to distinguish from human-produced work.
So the labels can demand new rules, but without enforcement tools, the rules don’t mean much. The real test is whether chart companies and streaming platforms will build or buy the detection systems needed to make this stick. Right now, that capability simply doesn’t exist at the scale required.
