I’m sure that most people have heard the term “deepfake”. Deepfake technology utilizes deep learning and artificial intelligence to create videos where people appear to say things they haven’t said. For an example, see this deepfake video featuring Barack Obama:
Clickbank Affiliate Marketing
You can see why this could be extremely dangerous. Notice how natural everything looks. The way his lips move and his facial movements all look completely natural and it is almost impossible to spot any imperfections that would make anyone think the video is fake.
Deepfake detectors have been pretty successful in identifying deepfake videos but the software requires training from hundreds of videos. So far, only a couple deepfake videos have succeeded in bypassing detector software. Researchers have found that subjects in deepfake videos rarely blink and have also found anomalies in the light reflections from the eyeball can be detected by AI. The problem though is AI is used to create these videos and AI is being used to detect them so the deepfake software will just get updated to overcome anomalies that are found by detection software.
Just imagine how this technology could be used to discredit public figures by making them appear to make hateful or racist statements or by impersonating close friends or family members of the public figures.
People are spending more time that ever online where these videos could easily be shared or sold in an instant. Once the video goes viral, the damage is done. It could easily take days for any detection software to find out if the video is a deepfake.