What prevents AI scrapers from continuing to scrape sites that contain a <Canary> tag but not follow the bad links?