But the "fail-dangerous" issue is more of a user responsibility, and I don't see that being solved anytime before the above technology is discovered (if ever).
Putting speculation aside, we know that AM is fail-dangerous. For all current purposes, we have better alternatives. For long-endurance power sources we have fission. For weapons we have fusion.
I cannot personally think of a common use where the statistical certainty of failure outweighs the risk.
My knowledge in this domain is quite limited, but maybe someday somebody will figure something out. Just the thought of antimatter existing is very cool.
http://en.wikipedia.org/wiki/Ivy_Mike
The Tsar Bomb test was restricted to 50Mt by missing out the 3rd stage - so it was primarily a fusion device. Had the depleted uranium tamper been in place it would have given the 100Mt the design was capable of - but with horrific fall-out.