Do people really audit such code? I doubt it. Does the code really not insert any additional code that allows bypassing the whole scheme (esp. If the point is to dynamically insert content).
I also think most of the time, the biggest threat is not the vendor being intentionally evil but the vendor making a mistake that leads to XSS which someone else exploits. After all, if the vendor is intentionally being malicious they can probably sneakily bypass this sort of thing.
