I have conflicting intuitions on this. In trying to sort out which of those intuitions merits more attention, I thought of a helpful analogy: normative powers of institutions work like spellcheck. People program spellcheck to accept certain words and underline words that are not already recognized in the dictionary the programmers use. Generally, when spellcheck underlines a word, I have a reason to change my spelling. But the reason to change the spelling is not because spellcheck said so-- it is because spellcheck is in sync with the dictionary used by the programmer, and the dictionary is authoritative (by democratic consensus? convention? let's just grant that it is for the sake of the illustration). Other times, spellcheck will highlight words it does not recognize but that I am licensed to use while writing a philosophy paper (like "noncontractual") or anglicizing a Greek word (like "nomos"). In other words, my compliance with spellcheck is optional because its normative force is only derivative, not original, and there are spheres in which its source is not authoritative-- such as philosophical articles in which Kant scholars, not Merriam-Webster, might have the last word.
This is not to say that spellcheck is redundant or impotent; it can give me additional reasons to change my spelling. If someone editing my paper knows that I typed it on a program equipped with spellcheck, she is likely to be more annoyed at my misspellings than she would have been had I been writing by hand because I had the resource available to make the correction myself. I might have new responsibilities in light of the new information. But I should have spelled it such-and-such a way all along. Whether discharging that duty to spell correctly was feasible for me prior to spellcheck or not is another question. But it may be an important one. If "ought" implies "can," maybe institutions and autocorrect and spellcheck broaden our possibilities-- our "cans"-- and thus broaden the "oughts" that apply to us.