AI recreations of real people, which clone their faces, distinctive voices and all, have become so convincing that even top computer specialists are finding it hard to tell the difference between fakery and reality. Creative "deep fakes" aren't all bad - they can bring dead artists and historical figures alive, to energize museums and schools - but most uses so far have been evil: Think fake pornographic videos of your sister or daughter, or scams that bilk people out of millions of dollars. Our guest, law professor Danielle Citron at the University of Virginia, has been advising companies and government officials on how to protect the public from the terrible harm that deep fakes can cause, but executives like Elon Musk don't want to listen. Her worst fears haven't come true yet - such as a deep fake that changes an election or triggers a war - but it's hard not to expect them.