I suspect that Anthropic, being public, and under scrutiny, will get sued into compliance from all over the world. It's not a foregone conclusion that the data they trained on remains fair game over the long term.
YouTube has a massive system in place to find and block pirated material. I can't remember the last time I saw a simple video rip on YouTube.
Nowadays you're getting gray area stuff like new AI generated stories based in the Harry Potter universe, which is probably copyright violation but it's not actually pirated video, so Content ID doesn't spot it.