I don’t think it’s that far away. It’s not just that technology has been accelerating over the past few decades, it’s that the humans have been gaining more and more knowledge about the bigger fundamental patterns that technology uses.
As a very rough example: if their demographic is Americans they will come out the gate with proper spelling, grammar and sentence structure that’s been run through LLM(s), and an overall branding that does not raise any immediate red flags. If someone tries to implement a blacklist or whitelist of words, they will walk past it like it wasn’t even there.
Moderation tooling is not as elaborate as we would hope, and all your work is with humans - who keep adapting.