There is a difference between censorship and propaganda. We were talking about the latter.

There is also a difference between government-mandated censorship and self-censorship. ChatGPT is almost exclusively doing the latter in order to avoid civil lawsuits, not the government busting down the doors. That’s obviously not even remotely the same as a Chinese LLM cracking down on Winnie the Pooh, because the God Emperor has a fragile ego.

Then there is the whole matter of training material. You won’t get most LLMs - including entirely open source ones with no commercial interest behind them - to spout your fringe political opinions as facts, because there is very little training material out there that agrees with you (or talks about how to launch an armed resistance - how many books and websites do you think exist on this topic?). A flawed democracy is still a democracy - and no serious scholar on this topic will call the US anything but that or variations of the term. Whether or not this remains the case after another four years of Trump is an entirely different matter. It’s not unlikely that the country becomes a hybrid regime like Hungary or worse, but an LLM that has difficulties with answering two questions about the past or present without hallucinating once can’t look into the future.

What annoys me the most about your comment is not that nearly everything about it is factually wrong, but that it’s nothing but whatsboutism, an attempt at defending what the Chinese regime is doing. That’s not a good look.