I was recently listening to music on Spotify and a song started playing that suspiciously reminded me of one by Lana Del Rey. The song is ‘Colour of Sense’ and the artist, a certain “woming pai music.” The song has more than 34,000 listens but it has all the appearance of having been created with artificial intelligence, basically because Lana’s voice and the melody of ‘Young & Beautiful’ are easily recognizable, although manipulated.
The most worrying thing is that Spotify’s algorithm has included it in my streaming queue, when I do my best not to consume AI-generated music that so obviously copies original recordings, by Lana Del Rey no less.
According to Last.fm, the song I listened to before that was ‘Almost a Lover’ by BEA1991. That is to say, “woming pai music” seems to have sneaked into the recommendation ecosystem of BEA1991, and probably other similar artists, imitating their sound and emotional patterns to fool the algorithm and, in the process, the listeners. Speaking of listeners, this “artist” has almost 15,000.
Not to mention ambient and relaxing or sleeping music playlists, which are completely full of sounds generated by AI, which go unnoticed due to the nature of this musical style.
The problem is starting to become really worrying, as fake artists like The Velvet Sundown become popular. The latest name to do so is Sienna Rose, a supposed soul promise who has accumulated millions of views with songs like ‘Into the Blue’, but who has received complaints for having been generated by AI, and everything indicates that it is.
We are not talking about “slop” in this case, but rather about extremely realistic creations that are confused with human beings. The public is beginning to generate tolerance for this type of products and assume them as another option: you just have to remember the way in which Bad Gyal praised his duet with Bad Bunny made with AI. He said it was a “great deal”: it was true. But do we want it on a platform that we use under subscription?
As reported by Stereogum, Deezer has reported that 50,000 songs published on its platform per day are created with artificial intelligence, 34% of all its new music. Spotify, for its part, has promised to get rid of the “AI-generated” garbage – again, the so-called “slop”, which in English means something like “sloppy” – but its efforts seem in vain, since listeners continue to come across songs like the one I quote, without wanting to. It is true that Sptify allows the option to “report” misleading content, but fake songs continue to proliferate. At worst, we are talking about real musical garbage floating around the platforms as if nothing had happened.
Now Bandcamp, one of the platforms that treats its artists best, has released a ray of hope: from now on, music generated with artificial intelligence is prohibited on its service. And they not only refer to music “completely” generated by AI, but also to music that has only been “partially” generated. They have a job, since the amount of music that is uploaded to Bandcamp daily is considerable. The platform has published clear guidelines for music made with AI:
1.- Music and audio generated wholly or substantially by AI are not allowed on Bandcamp.
2.- Any use of AI tools to impersonate other artists or styles is strictly prohibited, in accordance with our existing policies prohibiting impersonation and intellectual property infringement.
Bandcamp explains: “If you come across music or audio that appears to be entirely created or heavily dependent on generative AI, use our reporting tools to flag the content for our team to review. We reserve the right to remove any music suspected of having been generated by AI.
Bandcamp is not the most used streaming platform, nor is its streaming experience the most fluid, nor is its algorithm the most powerful, but it is an oasis for artists and listeners looking to consume music more ethically (and buy it directly if they wish). Now, they are taking a firm step by banning AI from their service, and I hope other platforms follow suit.
