A deepfake video of Tom Cruise took the internet by storm this week garnering more than 11 million views. In the video, Tom Cruise was golfing, doing magic tricks, and telling jokes. While harmless, these deepfake videos can have serious implications in the future.
Deepfakes use deep learning technology, a branch of machine learning that applies neural net simulation to massive data sets, to create a fake. Artificial intelligence effectively learns what a source face looks like at different angles in order to transpose the face onto a target, usually an actor, as if it were a mask. In a report, the U.S. Government Accountability Office (GAO) says the underlying technology of deepfakes “can replace faces, manipulate facial expressions, synthesize faces, and synthesize speech.” To a casual observer, this video will look authentic. If you start to pay closer attention, you can see small flaws in the recreation of his voice, mannerisms which don’t quite fit the dialogue and a slightly different body type.
Unless you were told before watching the video that it’s fake or you clicked on a link which highlighted that it was a deepfake video, the likelihood is that many people will believe that it’s real. A deepfake video of Tom Cruise doing magic tricks or another more recent example, Donald Trump in Breaking Bad can seem like harmless fun. A deepfake was made of Barack Obama using an expletive in regards to Donald Trump. The video was convincing, it’s only downfall being Barack Obama’s voice. It sounded a little unconvincing. There is a very small leap between a harmless video and a revenge video. As the technology for deepfake videos improves, the likelihood that people will believe these videos are real, also increases.
The GAO “Science and Tech Spotlight” states that whilst deepfakes to have positive applications, they are generally not used that way. “While deepfakes have benign and legitimate applications in areas such as entertainment and commerce, they are commonly used for exploitation,” the GAO says. “Deepfake technology has reached the point where authenticity of a video is almost impossible to confirm as genuine,” Brandon Hoffman, Chief Information Security Officer “Media outlets do not want to be the unwitting participants in causing widespread panic… With deepfakes they are in a position where they have to decide, without any technology to help them confirm authenticity, whether or not to run a piece with video that could be fake,” Hoffman said.
Chris Ume, the creator of the video wants people to be more aware of what they’re watching and that it might not be real. “I always try to make funny content, but at the same time, when people see these videos they realise and will learn what’s possible within the next few years,” he said. “The tech is evolving rapidly and will get better and will become more accessible as time goes on. Twenty years ago you have Photoshop, you didn’t know about fake photos so they started editing photos, and now people realise – like photos – that videos can be misleading. So it’s important on my side to create awareness so people start thinking twice when they see similar videos,” he said. “You will always have people misusing techniques, so you have to think twice when you look at something.”
ARTICLE: PAUL MURDOCH
POLITICS EDITOR: CARSON CHOATE
PHOTO CREDITS: OBSERVER