Describing a relationship that is related via a third party?

In phsycology we call variables that correlate covariants, thus saying they vary together but not necessarily cause each other.

From M-W:

Covariant: varying with something else so as to preserve certain mathematical interrelations

This doesn't necessarily mean there's a common cause to both of them (c is causing both a and b), but it conveys the idea that, even though a and b are related, they aren't necessarily a cause for one another.

Also see Third-cause fallacy

Another word is confounded, although this one is pretty much restricted to Statistics and may have negative connotations. From the wikipedia:

"We say that X and Y are confounded by some other variable Z whenever Z is a cause of both X and Y."

We could call the relation between A and B to be coincidental/correlated and the one between C and A/B to be causal relationship. I would suggest you to look into Correlation does not imply causation to not just get some more insight into the subject matter but also the relationship label.