Children and pets are not the same as adults. Guns are also different from language models and image generators. A gun is a weapon, but a language model isn't.

Adding certain protections might be necessary for objects that can otherwise cause bodily harm to the user (e.g. gun safeties), but if you think that people must be prevented from accessing information because they are too stupid to properly evaluate it, then you might as well abolish democracy.

I am not doubting that people can evaluate information incorrectly. The issue is that nobody can do it in an unbiased way. The people doing the censorship don't know all that much better and often don't have the right intentions, as is often demonstrated.

It has been shown that ChatGPT has strong political biases as a result of the tampering applied to make it "safe". I find this concerning.