The first examples are what is called the epistemic use of the modals "can" and "must": they're saying things about the speaker's knowledge and expectations, not about possibilities or powers in the real world. They both say "I am sure that" or "I conclude that" it isn't broken. Certainly can't is more likely there in BrE: I don't know about AmE.
In BrE, epistemic "mustn't" is more commonly used to mean something like "I conclude that", often with a hint of surprise, eg he mustn't have found the box means something like "I thought he would find the box, but judging from what happened afterwards, he didn't find it". It musn't be broken is certainly possible, but has a hint of surprise which is not there in It can't be broken.
The other use is deontic: it's not about the speaker's knowledge, but about potentialities and obligations in the real world. In that context, "must" and "can" have very different meanings: you must not even touch it means "you are not permitted to touch it", or "you risk some awful consequence if you touch it", whereas "you cannot even touch it (note that "cannot" is one word, not two) means "it is not possible for you to touch it". This can sometimes be intepreted as social or institutional possibility, and so means the same as "must not", but it could also refer to some other kind or possibility, such as "it is not physically possible for you to touch it".
The relationship between epistemic can and must has little in common with the relationship between deontic can and must.