To train LLMs it seems asinine to try to scrape things instead of making a clone. The only automated searches I could envisage to MAYBE be happening are those

To train LLMs it seems asinine to try to scrape things instead of making a clone. The only automated searches I could envisage to MAYBE be happening are those looking for vulnerabilities, and there are probably only a few dozen actors doing it. State actors seem actually more likely to use clones of everything.