I don't like the "is" because the assignment operator never is one of those things. I already provided my own variation that avoids that phrasing, but maybe it's inferior in some other way.

Yes, I see what you mean, and I agree. Something like "would" is really what we'd need here.

Merging the two suggestions, how about:

A using-declaration in a derived class C that brings in an assignment operator from a base class never suppresses the implicit declaration of an assignment operator of C, even if the base class assignment operator would be a copy or move assignment operator if declared as a member of C.

Looks fine, except that "brings in" is a reversion from "names" in P1787R6, which says that using-declarations do not copy-paste declarations around. (Admittedly, this is (now) a note, but I don't want to encourage the old way of thinking.)

Ah, I see. How about this then:

A using-declaration in a derived class C that names an assignment operator from a base class never suppresses the implicit declaration of an assignment operator of C, even if the base class assignment operator would be a copy or move assignment operator if declared as a member of C.

You mean that "from a base class" is redundant? If so, then yes, that's true, but in a note this seems helpful explanation, since it's exactly the puzzle solving that would be needed in the absence of this explanation that the note tries to spell out. (We could also say "an assignment operator (from a base class)", or "an assignment operator, which must be from a base class". But we shan't.)

Fixed

@opensdh: I know you were already OK with this, but could you take a final look? I appreciate your eye for avoiding ambiguity and redundancy!

No new opinion here. If someone has a brilliant rephrasing idea, that can be a separate change later that doesn't have the burden of "could".

OK, merging!

To recap/abstract the interesting point of the recent discussion: A phrase "a thing X that is a thing Y" can have two purposes: it can either constrain the set of possible Xs to those that are also Ys, or it can merely point out that an X is also always a Y, but both the property X and Y are importent to the context. In normative wording we might want to avoid the second construction, but in normal English I find it perfectly reasonable, and so it is also reasonable in this note.

"I touched the cold snow" doesn't mean I carefully selected the particular kind of snow that I picked up, but simply that my hands were getting cold.