Sign In

Communications of the ACM

ACM News

Facebook Researchers Say They Can Detect Deepfakes And Where They Came From


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Examining a deepfake image of former U.S. president Barack Obama.

Researchers at Facebook say they have developed a form of artificial intelligence that determine whether an image is a deepfake, and if so, track it back to its origin.

Facebook researchers say they've developed artificial intelligence that can identify so-called "deepfakes" and track their origin by using reverse engineering.

Deepfakes are altered photos, videos, and still images that use artificial intelligence to appear like the real thing. They've become increasingly realistic in recent years, making it harder to detect the real from the fake with just the naked eye.

The technological advances for deepfake productions have concerned experts that warn these fake images can be used by malicious actors to spread misinformation.

Examples of deepfake videos that used the likeness of Tom Cruise, Former President Barack Obama, and House Speaker Nancy Pelosi went viral and have shown the development of the technology over time.

From NPR
View Full Article

 


 

No entries found