The battle against deepfakes
Emailed on January 17th 2020 in The Friday Forward
I've written about deepfakes and how scary they are for our future, but if you're unfamiliar: Deepfake (n.)—persuasive yet synthetic media generated by AI depicting a real person’s likeness.
From harmless viral videos to malicious misinformation campaigns, deepfakes have infiltrated the mainstream.
Now online platforms are fighting back. Reddit said Thursday it will ban impersonation, including deepfakes (but not satire). Facebook announced a deepfake ban last Monday, but critics say there are loopholes wide enough to drive a bus through. Alas, it's a start.
Deepfakes are also top of mind for governments. AB 1903, a California bill introduced last Wednesday, would criminalize the use of deepfakes for wrongful purposes.
Even as some companies try to limit the technology’s use on their platforms, others are productizing it (this is how Black Mirror becomes reality). Snap recently launched Cameos, which stitches users’ faces into short videos. And last week we learned TikTok parent ByteDance developed deepfake face-swapping technology.