Well, I don't know, I've been making fun of Yudkowsky's positions in this comment section, but I think the official corporate position of the Machine Intelligen

Well, I don't know, I've been making fun of Yudkowsky's positions in this comment section, but I think the official corporate position of the Machine Intelligence Research Institute (MIRI, the institute dedicated to banning Research into Machine Intelligence) is that you defect in such a scenario in the modal case. As in it is morally correct and rational to defect, not just that they predict it would happen.