I suspect that you mean something like:

If there is an objective universal morality then agents converge on this universal at the limits of intelligence.

And thus perhaps paperclip maximizers have a tendency to become something else.