I wonder if we need legislation for AI training, at least for commercial use of training datasets for LLMs (and other generators like text to image), should be that authors of the training material should be Opt In rather than Opt out.

Most people publish their work for another humans, not for a corporation to use machine learning to do copyright laundering.

Just because you put something in robots.txt doesn't mean they will adhere to it.

The title is "how to ask..." not "how to force...".