Pretty sure that’s already decided. Publicly played movies and music are not available to be used. Why would the same not apply to posted images?

What court case set the president that you can’t train a neural network on publicly posted movies and audio?

I'd assume the precedent would be about sharing encoder data, which would be covered in bittorrent cases.

"Training a neural network" is an implementation detail. These companies accessed millions of copyrighted works, encoded them such that the copyright was unenforcable, then sell the output of that transformation.

Not being able to reproduce the inputs (each image is contributing single bytes to the neural network) is relevant. Torrent files are a means to exactly reproduce their inputs. Diffusion models are trained to not reproduce their inputs, nor do they have the means to.

If you post a song on your website and I listen to it am I violating your copyright?

If my parrot recites your song after hearing my alleged infringement, I record its performance and post it on YouTube is that infringement?

Last one, if I use the song from your website to train an song recognition AI is that infringement?

If I host a song I don't have license to on my website I'm violating copyright by distributing it to you when you listen on my site.

If my parrot recites your song after hearing it and I record that and upload to YouTube. I've violated your copyright.

If a big company does the same(runs the song through a non-human process, then sells the output) I believe they're blatantly infringing copyright.

Big Company is not distributing the input images by distributing the neural network. There is no way to extract even a single input image out of a diffusion model.

I should have specified the OP has legal rights to the song and the end user listening was under the same granted/implied license as a program doing the web harvesting, my bad.