HarmonyCloak is a concept for combating generative AI music.

"At its core, HarmonyCloak functions by introducing imperceptible, error-minimizing noise into musical compositions. While the music sounds exactly the same to human listeners, the embedded noise confounds AI models, making the music unlearnable and thus protecting it from being replicated or mimicked. "

It's kind of a poison pill for AI Music. The examples on the page are pretty convincing to the idea - you, a human being, do not notice anything in the "Cloaked Music" snippets but an AI trained on it produces less than acceptable results. It's a fun idea to think on and fascinating samples to listen to.

There have been other projects tackling similar problems with similar tactics. You may recall Nightshade was meant to introduce imperceptible noise to images that would make it impossible for an AI to train on the data and give good results. Again, fascinating to read about but I'm unclear on the adoption rates.

The core issue at play here is copyrighted material being used to train models with zero compensation to the original creators. Wanting to combat that from the inside is a sensible path to explore.