I had an understanding that quantum entanglement can be used to for encryption and guaranty there are no man-in-the-middle attacks by making sure the entanglement is not broken. if we can’t know the collapse happened how can we make such claim (mitm can meassure stolen briefcase put it back and no one would know)

This is incorrect. Quantum key distribution ("quantum cryptography" is a misnomer) is vulnerable to man-in-the-middle attacks. Security relies on having a pre-shared key which makes it essentially useless in practice (just use AES if you have a PSK).

AES 256 only requires millions of qubits to crack. What do we do once we think that's attainable behind close doors? It'd be better to be prepared.

https://www.fierceelectronics.com/electronics/aes-256-joins-...

This is a wildly optimistic estimate. Even if we had an error corrected quantum computer that could evaluate an AES key (complete science fiction for the forseeable future), running Grover's algorithm would take millenia assuming extremely fast gate times (1 ps).

In any case, doubling the key length would be infinitely simpler than using QKD.