I'm not sure I got your point... do you mean counter example because it proves it is not deontological or because it proves it is not desirable to be deontological? If you look only the consequences paul and mary actions are the same, they kill one to save two.
I think it is interesting to think about the practical consequences for the movement of the different approaches. If what matters is only extension it would justify to get adherents no matter what (i don't know, like using strong images, manipulated data...), on the other hand, from an intensional point of view you want adherents that get the point reasonably enough to push it further.
Maybe we also need a distinction between adherents that are only interested in knowing good charities to donate to, and those who are willing to push the ideas further
I wonder if the 'intensionality' generates a deontological framing. I mean, if Paul donates to AMF because he had malaria as an infant and Mary donates to the same charity because she thinks it is effective, then, Paul is not an EA and Mary is, right?