It's relatively trivial to photoshop misinformation in a really powerful and undetectable way- but I don't see (legitimate) instances of groundbreaking news over a fake photo of the president or a CEO etc doing something nefarious. Why is AI different just because it's audio/video?