285
Congress Wants Tech Companies to Pay Up for AI Training Data
(www.wired.com)
This is a most excellent place for technology news and articles.
Not a great comparison. AI training is not problematic because of consumption, it's a problem because it is then used to circumvent copyright law.
If you do want to argue with technicalities, you also have to contend with the fact that a large part of the concerned usage is not really free. Much of this online data is funded by advertisement. Scraping it constitutes circumvention.
There are no contracts signed agreeing to that exchange by either me nor the scrapers. Legally, it's free.