I'm not sure how you think safety problems could be fixed *before* they are discovered.
Through defensive design. By requiring that system design promotes safety; therefore, there are unlikely to be serious safety issues. The key is to design systems that are anti-fragile, AND that are robust such that random safety issues aren't emerging after product release.
Not only is fixing safety problems *after* they are discovered reasonable, it's really the only possibility.
This is not a valid excuse for designing and releasing or using systems with inherent vulnerabilities that are therefore likely to have safety impacting issues later, and therefore: incurring this extra liability.
It is useful to people who refrain from doing certain things they might benefit from because they are *not* 100% safe.
As you mentioned.... the benefit is minimal or uncertain, BUT the risk is real.
Everything that there is a risk of happening, eventually happens given enough time!
Most people drive their car less than 2 hours a day, BUT rely on their smoke detector to help protect their lives 10+ hours a day.
The fact that your automobile is very dangerous, is no reason for engaging in reckless behavior, in other areas: however --- it just adds to the probability of random death.
You are about 14 times as likely to die in a car accident as a house fire. Every day you drive a car is as dangerous as going 14 days without any smoke detector at all.
This assumes you are an average driver. But perhaps I am a far-safer-than-average driver driving a far-safer-than-average car.
Maybe your roads aren't as dangerous as the average. There is plenty of room for outliers here.
You can't possibly be sure that you are X times as likely to die in a car accident.
You claim to have calculated risks which are actually impossible for you to have calculated, which is the reason, that I know your claim about the relative likelihood is definitely false (That which cannot be true due to an absolute condition, is guaranteed to be false.).
Maybe you will die in a car accident and I can say "I told you so". It's all about balancing risk vs. reward. I want my house nice and toasty when I come home.
The Idea "I want my house to be nice and toasty when I come home; even if there is this substantial chance that some Chinese hacker can kill me"; is a bit of a depraved notion. The fact is it's not possible to calculate the "risk" part of risk/reward. The fact is any danger of incalculable risk is not worth it, if the danger is great enough. The reward has to be such that: the absence of the reward is as bad or nearly as bad as the maximum potential negative impact of potential hazards that may exist.