There is a word doing a lot of heavy lifting in music right now, and the word is slop.
It gets used for anything made with a generative model. It gets used before the track has been played. It gets used as a way of settling an argument without having one. And it has become the default reaction to a category of music that, by any honest measure, a lot of people are enjoying.
We are going to make the opposite case. Not because the criticism is baseless, because some of it is completely correct, but because the criticism has been aimed at the wrong target.
What the critics get right
Let us not be precious about this. There are real problems, and pretending otherwise would be insulting.
There is a genuine flood. On Deezer, AI generated tracks went from around 1% of new weekly releases in early 2024 to more than 40% by late 2025. Spotify says it removed roughly 75 million spam tracks in a single twelve month period. That is not a rounding error. That is a firehose pointed at a system that was never built to filter it.
There is genuine impersonation. Models trained to imitate a specific artist's voice, then used to put words in their mouth, is not a grey area. It is a straightforward wrong and it deserves the anger it gets.
And there is a genuine dilution problem. Streaming royalties come out of a shared pot. Every track uploaded to farm fractions of a penny takes a slightly larger slice from everyone else. If you are a working musician watching a million auto generated tracks appear next to yours in the same royalty pool, your frustration is not irrational. It is arithmetic.
So when someone says the flood is a problem, they are right. Hold onto that, because it matters for the next part.
What the critics get wrong
Here is where it falls apart. Every one of those problems is a problem of distribution incentives. None of them is a problem with the tool.
The flood exists because streaming pays per play. When a platform pays a fraction of a penny per stream, the rational strategy for a bad actor is volume. Upload ten thousand tracks, catch a few thousand accidental plays each, collect. That strategy existed long before generative models. AI did not invent it. AI just made the marginal cost of another upload approximately zero, and a system that was already gameable became trivially gameable.
Point the same tool at a different system and the behaviour changes completely. Nobody spams a storefront. There is no reward for listing ten thousand tracks that nobody buys, because a listing that does not sell earns you exactly nothing and costs you the effort of making it. A store does not pay out for volume. It pays out when someone hears something, wants it, and decides it is worth their money.
That is a different incentive, and it produces different behaviour from the same technology.
Slop, or drop
In dance music, the drop is the part you wait for. It is the moment the track stops promising and starts delivering. It is also, notably, a thing you can only judge by listening.
Slop is the opposite kind of word. It is a judgement you can make without listening, because it is a judgement about how something was made rather than about what it does to you.
We would rather use the first kind of word.
Here is the test we care about, and it is the only test that has ever actually mattered: someone hears fifteen seconds of your track and decides they want the whole thing badly enough to pay for it. That is not a test you can pass with volume. It is not a test you can game. It does not care what software was open when the track was made, and it never has.
Sample based hip hop was not real music. Acid house was a machine fault someone kept. Autotune was cheating. Drum machines were putting drummers out of work, and that one was literally the argument, in print, with union backing. Every single one of those was going to be the end of something, and every single one of them ended up being the start of something instead.
The list of tools that were going to ruin music is long, and music is still here.
The part nobody mentions
Here is a figure that reframes this entire argument. Around 93% of AI generated tracks on streaming platforms get fewer than 1,000 plays.
Spotify's monetisation threshold is 1,000 streams in twelve months. Below that, a track earns nothing at all.
So the overwhelming majority of the AI music people are angry about is not taking money from anyone. It is sitting at zero. The flood is real, the frustration is understandable, and the economic damage from the long tail of it is close to imaginary, because that long tail was never being paid in the first place.
Which raises an obvious question. If streaming pays nothing for the vast majority of this music, why is anybody uploading it there?
The honest answer is habit. Streaming is where music goes, so that is where people put it, even when the destination has no way to pay them and no interest in doing so.
What we are doing instead
Audibase is a shop, not a streaming service. Someone plays a preview, decides they want it, and buys the file. You keep up to 97% of what they pay, and it lands in your account at the moment of sale.
We are not going to pretend AI had nothing to do with it. We are not adding a badge of shame to tracks made with Suno or anything else. If it moves someone enough that they pay for it, it earned that, and how it was made is a production credit rather than a confession.
What we do care about, and will keep caring about, is the honest part. Do not impersonate a real artist. Do not sell something you do not have the rights to. Make sure your generator's terms actually let you sell the output, which for most of them means being on a paid plan when you made it. Those are real rules and we will enforce them.
Beyond that, we are not interested in policing how you made it.
The bit that decides it
The argument about whether AI music is real music will resolve itself the way these arguments always do, which is that nobody will remember having it. In ten years there will be tracks people love, and almost none of those people will know or care what the production chain looked like.
What we would rather do in the meantime is build the thing that pays you when someone loves it.
Slop is what you call it when it does not land. Drop is what you call it when it does. That is a judgement for listeners, and the only honest way to collect it is to put the track in front of them and see whether anyone reaches for their wallet.
Start selling your tracks, or read how it works for music made with Suno.