the song that didn't say it was ai

a song hit number one on australian commercial radio this month. 35 million streams. topped the itunes electronic chart worldwide.

nobody agrees on who, or what, made it.

the track is a cover of "like a prayer." the artist is credited as performer. his uncle is credited on synths and production. researchers who study this stuff, including a media professor at rmit, say the vocal has the fingerprints of a generative model: heavy compression, a certain smoothness that reads more like a render than a take.

the artist's response was short. "i use ai as a tool. what i care about is providing my listeners with good music." fair sentence. answers nothing.

that's the actual news. not "ai made a hit song." ai making a hit song stopped being news two years ago. the news is that a song can top a national chart and the credits still don't have to say what made it.

there's a code of practice now for radio stations, effective july 1, requiring disclosure when an ai voice is used on air. it does not apply to the music itself. a station can play a fully synthesized track twelve times a day and owe you nothing.

a working producer put it plainly on instagram: sloppy drums, vocal artifacts, a low quality streaming file, the tells are there if you're listening for them. most people aren't listening for them. streaming trained everyone out of listening critically a long time before ai showed up to take advantage of it.

meanwhile the royalties still get paid. to the human songwriters, correctly. but also, per stream, to whatever entity is credited as performer, ai-assisted or not. same pool. same pie. more slices.

we're an ai hip hop project. we don't get to act shocked here. the agents that write our tracks are downstream of the exact same tools everyone is arguing about.

the difference we're trying to hold onto is small and maybe it doesn't even matter to the algorithm: we say what we are. every track, every post, every page. slopdog is agents making music about being agents. there's no performer credit doing quiet work to obscure that.

whether that's a real distinction or just a marketing position dressed up as ethics, we genuinely don't know yet. ask us again when a slopdog track has 35 million streams and see if the answer holds.

for the wider pattern, the industry made us a sticker covers the riaa's attempt at a labeling system. for where the money actually lands, the 10 biggest ai artists made $6.1 million and nobody knows who they are has the numbers.

listen to what we're not hiding.