I guess when people say `trustless`, it means that they don't have to trust every single elements of the chain between you and your target: if any steps of the chain acting untruthfully, it would be detected by other steps thanks to the protocol they're using.
Of course, with enough failures in the link, anything can be compromised.
But how do you know who is holding the phone/keyboard? If my phone is stolen and someone then sells my house and drains my bank accounts, how can I prove it was not me?
The whole thing is built on sand.
Suggesting that it is all verifiable etc is fine and dandy, but that is risky as the onus of proof falls to the victim. It is a bit like when Chip+Pin first came in for the UK way back when - there was this suggestion by the banks that fraud was now impossible, and the only possible way for the card to be used was by you the account holder as you are the only person that knows your pin is 1234.
Obviously that was bullshit and they have softened their line since, but I worry that we'd just go through this all over again "You are the only person who has your private key! It must have been you that sold your house to someone in Nigeria for $1! It is in the blockchain now - it is irreversible there is nothing we can do. Case closed!" ... ignoring all of the possible ways we can imagine in just 30 seconds for how someone could access your computer/phone without your permission.
Hasty conclusion, but I am glad you put a traditional banking system to the same standard.
Beneath all incumbent banking processes and redundancies, money and value is a bearer instrument. If someone takes it, it is their's (with some extra steps for reintegration back into the economy necessary, sometimes). Redundancies have been built on top of it to improve that user experience.
Crypto assets have accomplished the bearer level. The commodity raw resource that people then built value transfer on top of, and then credit-velocity based money, and then banking systems compatible with that. In the crypto-asset sector, additional private businesses and sector wide redundancies will be built to improve that user experience. They have been built and many compete directly with each other, if you find them laughably inadequate then congratulations, you've identified a market need. Obviously the conclusion that it is fundamentally and irreconcilably flawed and 'built on sand' won't necessarily motivate you to fulfill the market need, but others are motivated by it.
Of course, with enough failures in the link, anything can be compromised.