I’d argue that’s already a mistake (I think I might have been your someone, unless you mean more recent).
All phylogenetic divisions are binary except in some odd cases, and we tend to want our taxonomy to reflect them. But there are lots of other biological categories. There are ecological categories like herbivore or pollinator, and broad morphological ones like tree or worm. There are more particular ones like ungulate or raptor, both of which were thought to be phylogenetic groups but turn out to have evolved more than once.
And then there are descriptors like slug or fish, used only for those parts of a phylogenetic group that share a particular property. Is it necessary to insist they be treated like they were only about taxonomy and so extended to the whole clade – and thereby made both quite different and redundant, since we already have words like gastropod and vertebrate if you don’t care about the form of the organism?
I’d argue this sacrifices the main point of terminology, which is to make it easier to talk about different concepts, of which clades are only one type. Saying “bony fish are a group of cartilaginous fish”, or “really vertebrates are a type of invertebrate”, might make sense if you only look at the evolutionary tree, but it also means you really aren’t interested in what the animals are actually like.
In any case, I just wanted to point out that either way, not all fish are sharks either. The bony fish and their tetrapod descendants are a sister group to Chondrichthyes, and sharks are only one part of that group; there are also the rays, and the more unusual deep sea chimaeras.
