I just saw this thread in random and have no idea behind the context of what AI dungeon is but the claims of active research for algorithms to detect "wrongthink" reminds me of another incident that I heard about long ago of an app called candid that was shilled as a free speech platform but was actually being used to structure the major tech firms anti hate speech algorithms (allegedly anyway).

If this is indeed being shilled on /mlp/ it might be a cause of concern if this is being used for the same purpose and if they have taken an interest in us for whatever reason. Though just as likely it could be mostly basis speculation and they are just interested in /mlp/ due to the success of 15.ai and are just censoring naughty stuff because they are a corporation, but I will try to investigate this if I get the time.

Just wanted to note as a semi public reminder and in case anybody else knew anything more.