Is there a little paradox here?

Mull briefly on this claim: Any argument which stems ultimately from induction, because it is based on generalizing from a limited observation rather than from the whole set of data, cannot be verified to be true. Is it self-referencing, self-refuting, defensible, or something else?