WTF is a deepfake and should you be worried?

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.

Amy Van Es

Today we’re learning all about deep fakes—the latest in a long list of inventions that we probably could have done without. But this new technology is a little different than, say, a revolutionary burrito holder. Deepfakes target our ability to tell the difference between what is true and false on the internet. And that is what we call a big fat problem.

What is a deep fake?

A deepfake is like photoshop for videos. It allows one person’s likeness to be transposed onto someone else. For example, you could think you’re watching a video of the President making an important announcement... but in reality, it’s a comedian making him say ridiculous things. Still confused? That’s fair. Check out this example of a deepfake to see the real deal.

How do deepfakes work?

Deepfakes use of a technology called artificial intelligence (AI), which is a computer that has been taught how to learn. In this case, the AI observes thousands of images and videos of the person you’d like to impersonate. Then, it uses all their different facial expressions and sound bytes to craft what is basically a digital puppet. Next, a different type of AI (called a decoder) layers that puppet with the original video of the impersonator. Et voila!

What are deepfakes used for?

A couple things.

Hollywood has been trying to get as close to “real” as possible when it comes to using special effects in movies. But deep fakes were actually invented by another film business: the porn industry. There’s a lot of money to make slapping celebrities' faces onto porn performers' bodies. Shocker.

But lately, deepfakes have made their way into the public eye as a tool to spread disinformation online. For example, a climate activist group based in Belgium altered a speech given by politician Shophie Wilmès using deepfake technology. In the modified video, Wilmès is seen giving a fictional speech about the link between the pandemic and the climate crisis.

How to spot a deep fake

So how do you spot a deep fake? Though this technology is new and still improving, there are ways to detect whether it’s real or fake. Here are some things to look out for:

  • Lack of emotion from the speaker
  • Unnatural eye movement
  • Slight blurring
  • Awkward body posture
  • Repeated movement
  • Monotonous voice

But it’s important to understand that this technology is advancing quickly. Eventually, there won’t be a way to tell the difference just by watching.

That’s… not great.

But here’s the good news: Deepfakes haven’t been as damaging to the information ecosystem as we thought they might be. But that’s because we already have plenty of cheaper, simpler tools to spread disinformation. It’s pretty complex to create a deepfake video. But a simple fabricated image (that you or I could create) can spread across social media just as effectively. Whether it’s a fake video or article or image, people generally don’t fact check everything they see online. Even worse? Most people don’t fact check the things that they share on their own pages.

Damn. So what can we do about it? 

Well, it’s important to be able to dig through the internet noise and find the original source of information. And one of the quickest ways to do that is to get off the page you’re looking at, and search from outside the organization. Often, if something is fake, you’re likely to find out from a quick Google.

But eventually, you’re going to come across some trickier fact-checking scenarios that require more than a search engine. And we’ve got a whole course to help you figure it out.

A Black man smiles at the camera.

Come here often? Get Memo in your inbox.