>>101375956If they can prove that the company obtained their works illegally, then they can face the same type of lawsuit that you or I could for torrenting a movie.
However using copyrighted material for profit is not inherently illegal. The fair use exemption is one such example that protects the creation of derivative works that are transformative determined using a four-pronged legal test.
But the AI algorithms are not even creating derivative works that contain elements of copyrighted material. They are creating new works trained off of data sets that include generalized infoemation ABOUT copyrighted material. There is a big distinction there.
If I were to create a parody animation about Luffy from One Piece and upload it to youtube, I would most likely be protected under fair use, especially if it were making fun of the anime.
If I wanted to write a story about pirates and included a character who was a skeleton man because I watched One Piece and liked Brooke, then that isn't copyright infringement at all so long as I didn't actually make that skeleton character a copy of Brooke.
The second example is what AI does. That's why it's called "machine learning". It learns associations in a similar way to how humans learn associations. What you're proposing is that no one can ever make media again after watching something that's copyrighted because the data set in their head contains copyrighted IP from someone else. That's absurd.