Deepfakes, combining artificial intelligence with machine learning, represent some of the most advanced technology for media manipulation in this digital age. Deepfakes create hyper-realistic images, audio, and videos that often make it look like a person said or did something they never actually did. While this technology has exciting potential for entertainment and education, its misuse poses significant risks to society.
Recently, one could hear a faked version of the voice of President Joe Biden in an automatically generated robocall that urged Democrats to not vote in the primary election. But some of the scariest facets of deepfakes relate to manipulating popular views and changing elections by viral on-site spreads over social networks. Even after a deepfake has been debunked, the damage is often done, as people may have already believed the misinformation. For example, Elon Musk is one of the most common celebrities used in deepfake scams, contributing to billions of dollars in fraud loss in the US. In addition to his wealth and entrepreneurship, Musk’s large media presence makes it easy for scammers to produce hyper realistic AI videos. Deepfakes, as they become more realistic, deplete public trust in media and digital content. Seeing is no longer believing. And there will likely be skepticism against even real videos as some sort of the “liar dividend,” whereby those in the hope of evading accountability would call an actual video fake.
Deepfakes symbolize both the marvel and menace of modern technology. As they further blur the thin line between reality and fiction, society will have to continue being vigilant in adapting to the challenges they throw up. Using innovation, regulation, and education in a mixed approach, we can utilize the benefits the technology brings while navigating its darker implications.
Links to videos that show deepfake examples: