Comment by Aachen

> an AI vendor will have a datacenter full of compute capacity. It feels like this solution has the problem backwards, effectively only limiting access to those without resources

Sure, if you ignore that humans click on one page and the problematic scrapers (not the normal search engine volume, but the level we see nowadays where misconfigured crawlers go insane on your site) are requesting many thousands to millions of times more pages per minute. So they'll need many many times the compute to continue hammering your site whereas a normal user can muster to load that one page from the search results that they were interested in