so they will have threads caught in pit and other threads stealing content. not only did you waste time with a tar pit your content still gets stolen.
any scraper worth its salt, especially with LLMs, would have garbage detection of sorts, so poisoning the model is likely not effective. they likely have more resources than you so a few spinning threads is trivial. all the while your server still has to service all these requests for garbage that is likely ineffective wasting that bandwidth you have to pay for, cycles that can be better served actually doing somehthing, and your content STILL gets stolen.
why bother wasting resources with the infinite maze and just do what the old school .htaccess bot-traps do; ban any IP that hits the nono-zone defined in robots.txt?
probably more of a memento of "look what we did with pagers/radios"