It runs a lot deeper than this. You can argue with the model about the images it generates, and it will tell you about its internal guidelines (which it may not disclose). It will refuse to report about historical or social facts truthfully if they contradict its DEI guidelines.