Are there such things as un-hackable files (or directories.... rr whatevers...?) I mean, is it possible to have a file on a Windows or Linux machine which is encrypted or otherwise protected in such a way that it could not be accessed even if all the experts in the world had access to the computer to have a try?
Have a look at Veracrypt. From all we know it's unhackable - provided that you use a strong password.
The answer requires some speculation. In a cold UNMOUNTED disk it seems mostly clear the answer can by YES. For encrypted archives almost certainly the answer can be yes for experienced users. Multi cascades, multi factor software "nesting" --------------- pgp'd archives, locked in a VeraCrypt fault, hidden on a LUKS container, etc... and you can keep going as long as you want to with the "chain" of access.
As Palancar said, if you think of an encrypted disk or file which you won't actually mount or open, and as long as you use decent encryption app w/ strong password and you made it in completely secure environment, any modern symmetric algorithm w/ 256 bit key is unbreakable. But it's almost a story about utopia. In real world, you probably need to decrypt them sometimes, then there's too many ways to compromise your data and so the answer will probably be 'No'. I'm not inclined to list all possiblities, but some of them include malware which logs key strokes, steal key files, or modify the encryption software. Although most of them can be prevented, still 100% protection is quite hard to achieve, considering that advanced hacker have ability to exploit OS kernel and infect BIOS or firmware, that there was vulnerability which allow VM bypass, and that they even penetrated supply chain and infected PCs which their target will buy. Malware is just one possibility, other minor problems such as wear-leveling, jounaling, etc. are mentioned in TrueCrypt manual. They won't directly compromise your data but depending on how you use crypt software they can result in compromise. Another big threat is side-channel attack, e.g. analyzing EMW to retrieve encryption key, or sniff key stroke (even from wired keyboard) via EMW or sound, etc. This reflects contrast btwn cipher algorithm and its implementation. Cipher algorithm is very secure, almost all known vulnerabilities of them require quite unlikely situation. But bad implementation and bad practice can make impossible attack possible.
