AI and deepfakes: The genie is out of the bottle

From Hollywood and Washington to social media timelines and the World Wide Web, deepfakes — a combination of ‘deep learning’ and ‘fake’ — and other synthetic media made with artificial intelligence (AI) have been proliferating quicker than any of us could have imagined.

In 2023 alone, at least 95,820 deepfake videos were found to have infiltrated the internet.

Sure, it may be cool and fun to watch Kendrick Lamar morph into Kanye, Will Smith, and Nipsey in the Pulitzer-prize-winning rapper’s 2022 single The Heart Part 5. You may even feel a sense of catharsis when replaying a rendered clip of Jon Snow apologising for the final season plot of Game of Thrones if you are among those (still) dissatisfied with how the iconic series ultimately ended.

But what happens when these instances of AI manipulation are applied for more nefarious purposes? Data and surveys indicate that most people cannot reliably detect deepfakes, with some admitting to watching videos they initially thought were real, only to find out later they were fake.

For the most part, the spotlight on the dangers that deepfakes pose typically falls on disinformation of the political variety. This technology, however, has been consistently used to harass and abuse women, including the likes of Taylor Swift, whose privilege and pop star status did not exclude her from being a victim of non-consensual pornographic material.

“Unfortunately, we should not be that surprised that deepfakes are used to degrade women,” assistant professor at Université Laval’s Department of Information and Communication Sophie Toupin tells TRT World.

After sexually explicit deepfakes of the Anti-Hero singer went viral on social media platforms in late January, Swift’s ardent fanbase, known as Swifties, sprung into action. Fans quickly swamped X with positive images of Swift and reported accounts that were sharing the deepfakes, causing the #ProtectTaylorSwift hashtag to trend.