You describe "decrypting" a hash, which doesn't make sense; a fully general First Preimage attack is by the Pigeonhole Principle a near-philosophical impossibility (barring an attack that somehow allows enumerating all preimages in something like shortlex order with efficient seeking, which would be an absurd and unreasonable criterion based on current understandings of cryptanalysis.) Would a regular, reliable, sub-1mo Second Preimage attack cause this question to resolve?

What about a sub-1mo Collision attack?

What if the attack in question is purely classical?

What if the attack in question is initially thought to be QC-dependent, but researchers later realize the algorithm could be run perfectly well on a purely classical computer?

There is no such thing as SHA-256 encryption. It is not an encryption algorithm, it is a cryptographic hash function. You cannot encrypt files with it.

There are certain properties a cryptographic hash function must satisfy. I'm going to assume this question is about violating those properties (which would lead cryptographers to consider the hash function "broken"), rather than anything to do with encrypting files.

To the best of our current knowledge, a quantum computer could provide only a quadratic speedup compared to a brute-force attack (using Grover's algorithm). That would take the attack from a 2^256 to 2^128, but even assuming an extremely powerful quantum computer this is still completely infeasible to compute.