Nudification of over 680.000 women by a social bot, impersonation scams worth millions of dollars, or spreading political misinformation through synthetic politicians are just the footfall of the rise of deep fakes. As every technology is simultaneously built with its counterpart to neutralize it, this is the perfect time to fortify our eyes with deep fake detectors. In this talk, I will introduce robust and accurate deep fake detectors by exhaustively analyzing heartbeats, PPG signals, eye vergence, and gaze movements of deep fake actors; including our renowned FakeCatcher. Moreover, we will exploit their heartbeats to trace their source generator, by unveiling residuals of different generative models. Key Takeaways: Original content has authenticity clues (i.e. human watermarks) that we can exploit to combat deep fakes. Source detection for deep fakes can reveal a brand new perspective for preserving "trust" in media. The detectors that we built on physiological, biological, and physical priors are ready to evolve our natural detectors -- our eyes -- against a deep fake dystopia.

Be the first to comment