Do you believe Mark Zuckerberg bragged about having control of billions of people’s stolen data? If you do, you feel for a deepfake.
First things first: what is a deepfake? Well, it’s the digital age’s answer to Photoshop. Deepfakes use an AI function called deep learning to create images or videos of fake events. Due to the advances in technology, deepfakes are getting harder to spot.
Needless to say, deepfakes have the potential to cause harm and mayhem. Here’s why they’re dangerous and how you can identify them.
The Purpose of Deepfakes
One 2019 study shows that 96% of deepfake videos are pornographic. Most of them work by mapping faces from celebrities onto porn stars.
Beyond porn, people can use deepfake technology for mischief. With new technologies, even unskilled people can make deepfakes with a handful of photos. This is the main reason why they present serious security threats.
The technology is also capable of creating convincing fictional photos from scratch. For example, “Maisy Kinsley” was a deepfake of a non-existent Bloomberg journalist. “Katie Jones” was likely a deepfake created for a spying operation.
How to Spot a Deepfake
Deepfakes used to be easy enough to spot, but this is no longer the case. As soon as experts find a weakness, it gets fixed.
In 2018, researchers found out that deepfake faces didn’t blink normally. Most images showed people with their eyes open, so the AI couldn’t process blinking. As soon as the findings got out, however, deepfakes with blinking started to appear.
The good news: we can still identify some signs of a deepfake. For instance, the skin tone might seem patchy, and the lip-syncing might be bad. You may notice flickering around the edges of faces, especially with poor-quality deepfakes.
Finer details such as hair are particularly hard to render well. Some styles, such as frizzy hair, are all but impossible to create. Poorly rendered teeth and jewelry can also be a solid indicator, as can inconsistent lighting effects.
Many deepfakes also look unnatural when slowed down. If you’re watching a video on a big screen, try slowing down the suspicious video’s playback. By zooming in and examining the images in detail, you may notice some discrepancies.
The Impact on the World
At this point, it’s hard to say what dangers the deepfakes present. The main threat seems to be creating a society where people can’t trust each other.
The security risks are obvious as well. For starters, some deepfakes may allow scammers to mimic biometric data. This is a major potential issue for systems that rely on voice, face, or vein recognition.
Of course, tricking humans tends to be even easier. Imagine your “mother” asking you to transfer money to her bank account via WhatsApp. For people who can’t identify deepfakes, this can be a frightening prospect.
More on the Deepfake Threat
As you can see, deepfakes are not something we can afford to ignore. Still, there are solutions. Many AI detection systems already allow us to spot a deepfake. Online ledgers have tamper-proof records, so any manipulations will stick out.
Want to learn more about the dangers of a deepfake? Interested in the latest robotics and AI trends? Keep reading our Technology section!
