It is true but not really relevant that they're smarter and more powerful than me. Since they don't trust me, they won't let me decrypt the disk D but will demand the key K to do it themselves. Recall K = D XOR P for some readable plaintext pictures or files, etc., I've already chosen. The "decryption" yields the perfectly readable output P = D XOR K.
Well, I read the original comment as being a: "here, K is the password, D XOR K is your results" idea -- based on that any D XOR K is technically a decryption, whether random stuff comes from the process.
If you intend K to provide "perfectly readable output", then
(1) if you actually produced the contents of D yourself to hide data, then you need to come up with some scheme so that K provides "perfectly readable output" and some alternative V provides D XOR V (or another decryption scheme) that gives you your actual secrets. I don't think it's that easy to have "perfectly readable data" on D XOR K plus have your secrets with another key. Except if you mean through steganography, but then K is not needed at all.
(2) if you were just send a random noise drive to "frame you" then you need access to the drive to come up with a K so that D XOR K decrypts to valid data.
This is a one-time pad. Given a ciphertext D it's trivial to come up with a key K that "decrypts" D to whatever plaintext P you want (K = D XOR P). This is why one-time pads are immune to brute-force search.
They'd ask you to provide actual readable files, pictures, etc, or else would reject your "decryption".