Has anyone made a kind of interactive post-Orwellian censorship sandbox? Seems like LLMs should make this possible. E.g., the user posts a trivially true thing

Has anyone made a kind of interactive post-Orwellian censorship sandbox? Seems like LLMs should make this possible. E.g., the user posts a trivially true thing and the LLM starts to reply with falsehoods, half-truths, concern trolling, etc. If the user can keep the truth above some percentage of the posts (or get the truth to bubble up to top comment) then they win that level. Maybe even different levels like Space Invaders so the LLM speeds up with each level, perhaps even adding more sock puppets each level. Would definitely be instructive. If done right it should also have an effect on the way the user consequently reads and assesses content on HN, Reddit, etc. :)