Photos does a lot of extra work on import (merging RAW+JPEG pairs, generating previews, database indexing, optional deletion), so my guess is a concurrency bug where a buffer gets reused or a file handle is closed before the copy finishes.
Rare, nondeterministic corruption fits the profile.
What's the point of it? It is well known in the industry they ignore bugreports.
Also, this bug doesn't affect the majority of users, so it won't ever be fixed.
I'll tell you a secret though that kind of pisses me off. If you have shipped with a bug, that automatically lowers the perceived priority as well. You know, as opposed to introducing a new bug in a new release. "We've already lived with that old bug…" seems to be the mind set. Oh well.
To be sure though, if you saw the number of bugs that queue up for a popular app like Photos, you'd know that fixing all of them is not going to be possible — some kind of system of prioritization is required.
This is unfortunately still true. I've had some radars de-prioritized with this exact reason and there are few things more infuriating.
It's really a sign of an overwhelmed, dysfunctional organization, but I heard from an ex-CoreOS guy that that's an intentional management choice, "people perform better under pressure"... (although I'm guessing Photos is not in the CoreOS org)