Deepfakes use computational techniques to transform photos, video, or audio so one person appears as another. Real-time deepfakes are now possible with open-source tools, meaning someone could hijack a conversation with a synthetic voice. The speaker's team demonstrated the technology by deepfaking his own likeness in a live presentation.
I can channel my grandpa now, like, "What are they gonna think of next?" So really, I would say deepfakes are using advanced computing algorithms or processing to render images of one person into appearing like an image of another person. And it can be photos, it can be videos, and it can be audio files or live audio. The real-time is really what the scary part that's emerged is, and open source and tools that are out there that you can actually do it in real time. So if I'm having a conversation with you and I don't actually know you, Bronwen, and you might think this is Aaron calling, but it's actually Aaron's deepfake, which unfortunately now exists based upon our research and my team deepfaking me in a tabletop and in client demos. And it's both cool and scary at the same time.
