You're ignoring the information about the source.

It's not about a heuristic on text of unknown provenance -- it's about publishers that exert a certain level of editorial control and quality verification. Or social reputation mechanisms that achieve the same.

That's what is preventing your "model collapse". Reputations of provenance. Not pure-text heuristics.

Anti-slop sampler and bigram/trigram stylometric analysis absolutely can give evidence that someone used a particular LLM with particular settings.

But in general it’s hard to separate human from AI text.

Unless you're OpenAI and provided like 98% of ai generated text that was posted to the internet.