It's two arguments.
Firstly, you've made the assertion that someone whose own life of at risk based their decisions is more likely to make safe decisions. As you say, it's an intuitive assumption to make, but reality tends to be very counterintuitive, especially when it comes to human logic and risk. Look at misconceptions around dangers of fighting vs driving, for example, or health risks and causes of death. So my first argument is that you should find evidence for that assertion before relying on it. I just don't trust it.
My second argument is that, even if it is true that people will take fewer risks if they know their life will be affected by those risks, I still don't think that would make that a useful tool in judging absolute risk. This is because human appetites for risk seem to vary a lot. Judging whether a building is safe based on whether the builder sleeps there is meaningless unless we know the builder's own propensity for risky decisions.
To give a more concrete example not directly related to risk: say you've got a disease, and I come to you with two medicines, A and B. I won't tell you how effective they are, but I will give you a double dose of medicine B, which will double its effectivity. Which one of these medicines should you take?
The logical answer here being that you just don't know, because you don't know the baseline effectivity of the medicines. Maybe they work about the same, in which case the double dose is obviously better. Maybe A is ten times more powerful to start with, so it doesn't really matter that B gets doubled.
Similarly, unless we have some indication of people's baseline propensity for risk, I don't think we can effectively make genuine (as in, statistically valid, data-backed) judgements about the situations you're describing, however counterintuitive that might feel.