Could a digital shield stop AI bots from scraping original music?

From American Public Media, this is Marketplace Tech.

I'm Megan McCarty Carino.

Last month, a band called The Velvet Sundown surged in popularity on Spotify, with its 70s

inflected rock and slightly too surreal publicity photos. Yeah, it was all generated by AI.

Synthetic music is getting better and better,

leaving a lot of human artists concerned that their original work

could be used to feed the machines that might take their jobs.

Now, some researchers are fighting back with a tool called Music Shield.

It basically inserts digital artifacts into original songs. They're

imperceptible to humans, but prevent AI bots from training on or editing these works.

Jin Liu is an associate professor at the University of Georgia, and he's currently working to

launch a beta version of the Music Shield in September.

So there is a perceptual gaps between humans and machines.

They may understand the music in different ways.

So the basic idea for MusicShall is that we use imperceptible noises to change the underlying

musical features of the music, which means that if you use our framework to protect your music,

when you listen to it, no difference.

They'll sound like your original music track.

However, if you feed this protected version to AI models,

regardless of the models to use this for training or editing, the extracted

underlying musical features will be totally different from your original features.

...