You can throttle APIs so they are still useful for a human browsing content, but essentially worthless for scraping meaningful amounts of data.
You can also detect accounts that are displaying scraping behavior and block them.
An LLM only has to download text once. You could easily coordinate a scrape across various IPs and accounts over the course of a couple of months and it would be really hard to detect. Especially if you target alternate GUI api based versions of the site.
Most users are not technical
Most technical users are still happy to use a GUI over writing their own scraper / visualizer