Here's a thought experiment: You are a hacker with a 100% (to your knowledge) secure notebook full of company secrets. I am an attacker and I have your son/daughter as a hostage. I am asking you to unlock that laptop for me in exchange for your kid. What do you do?
Do you really trust your failsafes enough to risk the life of your own kid? Or do you just unlock the goddamn notebook for me?
I mean, the right way to do it is with a precommitment to preventatively destroying the thing the attacker wants from you, via some kind of dead-man's-switch. You don't let your future self make the decision of whether to give up your secrets; your hypothetical future self is under duress. Your present self is not, and therefore knows better.
Of course, they still probably won't believe you that the secret is destroyed with no backup. But now you actually have no option to cooperate with them, so at least you're off the hook for the moral responsibility of whether the hostages live or not any more. The only ones that can make a decision that will causally influence whether the hostages live or die, at that point, are the hostage-takers.
That kind of dead-man-letter defense is a deterrent that relies on the attacker's ability to assess that the information can be destroyed, before going through the trouble of coming after you. If they are unable to assess it, or their intelligence is wrong, or you never let it wildly known that there is a dead-man-letter, once hostages are taken, you're still in that situation.
Even if you don't consider it as your moral responsibility, it won't bring dead people back to life. The cost is still the same. And if you survive, the trauma will still be there.
I mean at that point, you are better off acting insane.