This is one of the reasons I'm extremely unimpressed by complaints from openai and anthropic that other labs are "distilling" their models based on them... Basically: "You're training your model by running it again our own model which is itself a gargantuan copyright and content violation of a scale never seen before by humankind, HOW DARE YOU"

Whatever is necessary for the functioning of the (democratic) society. That would be legal.

btw, why isn't Greta Thunberg standing up in front of the UN delivering a "How dare you" speach about AI ?