How about this, then. It's my (possibly incorrect) understanding that all the big LLM products still lose money per query. So you get a Web request from some bot, and on the backend, you query the corresponding LLM, asking it to generate dummy website content. Worm's mouth, meet worm's tail.
(I'm proposing this tongue in cheek, mostly, but it seems like it might work.)
How about this, then. It's my (possibly incorrect) understanding that all the big LLM products still lose money per query. So you get a Web request from some bot, and on the backend, you query the corresponding LLM, asking it to generate dummy website content. Worm's mouth, meet worm's tail.
(I'm proposing this tongue in cheek, mostly, but it seems like it might work.)