With reference to exercise:111c378a55, more specifically this revision: https://www.stochiki.com/w/index.php?title=exercise:111c378a55&oldid=17182

When we specify claim frequency and claim size, "loss" refers to aggregate loss which is the sum of the claims.

A loss can be zero in an actuarial context. So when we say that we have a random sample of two losses, it doesn't assume the claim frequencies are positive.

I undid the edit and added some clarification and precision to the exercise.