If those IEEE journals require that copyright be signed over to them without granting permission to the authors to make available their work, I don't see how they can publish in them.
The policy on which the MIT faculty voted allows opt-out provided the author(s) provide a reason. The new policy is a great step for MIT and the research community as a whole but still requires MIT to stand up to journals who demand exclusive rights. It will be interesting to see how much backbone MIT asserts.
It's nice that it's MIT though since they're one of the institutions with the academic weight that a journal wouldn't want to cut themselves off from the authors there.
That said, I suspect "open access" is to the traditional world what P2P is to the music industry: it's scary and presents a real threat to their livelihood.