The problem is that you usually don't know what happened. Often the article lives on behind a paywall, and then you should (must!) use that one. To republish an archived copy from own repo will be a copyvio in many jurisdictions.

The interesting thing is (I guess) if we can quote the important fact in such a way that we don't make a copyvio yet make it possible to validate the value.

Usually you can quote a fragment and also republish that, but you can't copy the whole article and make that available for others.

Sorry for being such a pest and saying this! ;)

Jeblad