Spotify Deleted 75 Million Songs. It Should Have Deleted More.

Picture a river with seventy five million cigarette butts floating in it. Now picture someone finally wading in with a net. That is roughly what Spotify announced, as reported by Variety in September 2025 and echoed across the trade press. Over the previous twelve months the company removed more than seventy five million tracks it classified as spam, and it is rolling out a filter built to catch the next wave before it ever reaches a listener.

Some people heard that number and panicked about censorship. They have it exactly backward.

Read what they actually deleted

Spotify was not hunting AI music. It was hunting fraud. According to the same reporting, the spam filter targets a specific set of behaviors: mass uploads, duplicate tracks, search manipulation, and songs engineered to be just long enough to trigger a royalty payout before looping again. That last trick is the tell. A thirty one second nothing, uploaded ten thousand times under a hundred invented names and streamed by bots on repeat, is not music. It is a coin machine in a song costume, and every coin it pulls comes out of the same pool that pays real people. Deleting it is not silencing an artist. It is unplugging a slot machine that was rigged to drain everyone else's tips.

The distinction that matters

Here is the line Spotify was careful to draw, and it is the right one. Charlie Hellman, the company's global head of music product, put it plainly, telling Variety, "We're not here to punish artists for using AI authentically and responsibly." The target is behavior, not tools. An honest musician who used AI to help build a track and uploaded it once, like a human being, has nothing to fear from a filter that is looking for someone uploading by the thousand. To keep that line visible, Spotify is also backing an industry credits standard, developed through the body known as DDEX, that lets artists disclose exactly where and how AI touched a song. Transparency for the honest, a net for the fraud. That is the correct pairing, and it is almost exactly the arrangement I argued for when I wrote that a universal label should sit under an honest story.

Why this helps everyone, the AI curious included

A streaming catalog is a commons, and a commons dies from overgrazing. When the shelves fill with algorithmic filler, three things rot at once. Discovery breaks, because the recommendation engine is now digging through mountains of chaff to surface one real song. Trust breaks, because listeners stop believing that a name on a track means a person stood behind it. And the money breaks, because a finite royalty pool split among an infinite pile of garbage pays everyone in dust. Removing seventy five million spam tracks widens the shelf for actual songs, sharpens the recommendations that real artists depend on to be found, and returns a few of those siphoned pennies to the people who earned them. None of that wounds an honest artist who happens to use AI. All of it wounds the spam farm. That is the definition of a good policy.

The honest worry

No filter is perfect, and it would be naive to pretend otherwise. A system trained to spot spam will sometimes flag a small, strange, or unusually prolific human artist who looks, from a distance, a little like a bot. That is a genuine cost, and the answer is a fast and humane appeals process and a filter that keeps getting smarter, not a shrug. But the fix for an imperfect cleanup is a better cleanup, not leaving seventy five million pieces of junk in the water because a few real fish might get scooped by mistake. A city does not cancel garbage collection because the truck once grabbed a bin it should have left.

Do it again

The most striking thing about the number is not how big it is. It is that it piled up at all, quietly, while the doors stood open. Seventy five million is not a victory lap, it is a confession of how bad the flood had gotten before anyone moved. Good on Spotify for finally showing up with a net. The correct next headline is not that they went too far. It is that they went back in for the next seventy five million.