Google can still integrate web content into llms but putting it into a vector database, no training needed. I am pretty sure this is what they already do anyway.
In other words, outlawing web scraping for training data is no way around this. Everyone signed a deal with the devil long ago.