avatarLinda Caroll

Summary

The text discusses the realism and potential dangers of deepfake videos, focusing on a convincing fake Tom Cruise TikTok account and other examples.

Abstract

The article raises awareness about the increasing sophistication of deepfake videos, using the example of a TikTok account that features a highly realistic fake Tom Cruise. The account, which has millions of followers, showcases deepfake videos that have left viewers stunned and questioning their authenticity. The author explains how deepfake technology works, involving facial mapping and AI voice recreation, and highlights the potential dangers of this technology, including non-consensual pornography and political manipulation. The article ends with a reflection on the challenges of trust and discernment in a world where deepfakes can be indistinguishable from reality.

Opinions

  • Deepfake technology is becoming increasingly sophisticated and realistic, as evidenced by the deepfake Tom Cruise TikTok account.
  • The author expresses concern about the potential dangers of deepfake technology, including its use for non-consensual pornography and political manipulation.
  • The author suggests that the prevalence of deepfakes will make it increasingly difficult to discern what is real and what is fake.
  • The article implies that the development and use of deepfake technology may have unintended consequences and create a "f***ed up dystopia."
  • The author questions the ability of technology companies to effectively detect deepfakes, given the limitations of existing detection software.
  • The author suggests that the widespread use of deepfakes could erode trust and create a world where people can no longer trust their own eyes.
  • The article concludes with a reflection on the potential end of the world as we know it, quoting T.S. Eliot's poem "The Hollow Men."

This Is Not Really Tom Cruise. It’s AI. It’s Fake Tom Cruise. Can You Tell?

Some people refuse to believe this isn’t Tom Cruise. It’s too real. Do you know how easy it is to make deep fake videos like this?

screencap by author // source: #deeptomcruise

Say “AI” and you know what most people think of?

They think of ChatGPT, which can now pass AI detection more often than not while many humans don’t, and I wish I was kidding, but I’m not.

Or they think of one of the AI art programs that lets people create art with a few text prompts. Look at you, painting like Picasso!

Know what they don’t think about when you say AI?

Videos. Deep fake videos.

And we should. We really, really should.

Because they’re scary real.

Y‘all, this is not Tom Cruise...

It’s a deep fake Tom Cruise. From the TikTok account #deeptomcruise.

Can you even believe this?

I repeat. This. Is. Fake. It’s not Tom Cruise.

Viewers are stunned. Just stunned.

The creases in his forehead when he raises his eyebrows. The way he laughs and fixes his hair. Looks like Tom Cruise.

Except, it’s not.

Clip after clip, viewers are saying if this isn’t the real Tom Cruice, they’ll be stunned. A lot of viewers don’t believe it. Won’t believe it. They think it’s Tom Cruise and someone is trying to fool them into thinking it’s not.

Surely this has to be Tom Cruise. Right?

No. Sorry. It’s not.

screencap by author // source

If you also note the date on the comments, this was 2 years ago.

Two. Years. Ago.

How much do you think deep fake technology has improved in the two years since the deeptomcruise account was made?

Here’s another one. 28 MILLION views.

Does Tom Cruise speak Spanish? Who cares. Fake Tom Cruise does.

He speaks Spanish and laughs at himself. Swaps back and forth between English and Spanish and it’s so real it’s unbelievable.

One viewer said he “knows” it’s the real Tom Cruise because the teeth are off center, like Cruise’s teeth are. For him, that’s what means it’s actually real. Like, AI can’t fake that kind of detail, right?

Dude. Yes it can.

That’s how deep fake works.

The account has 5.2 million followers. Doesn’t matter which clip you watch, people ask if it’s real. It’s not. It’s fake.

Deep fakes are good. Really good.

Last one. He falls, he laughs, he makes a joke…

Again with the comments. People are stunned.

screencap by author // source

Disturbingly accurate. Is this really Tom Cruise? Insane. A bit scary how real it looks. [Comments from youtube, source]

That’s the kind of stuff people say. Rightfully so, I might add. Because it’s uncanny how accurate the fake is.

You know why, right?

How? HOW do they do this?

Basically, it’s facial mapping. All they need is a whole bunch of photos of the person you want to fake. The more photos the better.

Here’s what they did. They grabbed a whole bunch of photos of Tom Cruise and fed them into an AI machine learning application. Same way ChatGPT was “fed” words from the internet.

Not like there’s any shortage of photos of Tom Cruise, right?

Then the AI got to work “learning” how that Tom’s face moves, based on all the photos it was fed. What expressions does he make? What happens to his eyebrows when he smiles. It “learns” his expressions just like ChatGPT learns how words go together.

Then, all they need is a willing model. Doesn’t have to look anything like him. And they “map” Tom Cruise’s face onto another face.

And voila — deep fake. Have a look…

The second part is AI voice recreation.

Did you know that’s a thing?

There are AI programs out there that are as easy to use as ChatGPT. You create a login and make an AI of your own voice.

Each of them works slightly differently.

The most accurate ones ask you to read a whole lot. 70 or 100 sentences. Feed them back into the program and voila. The AI learns how you talk.

You can make your own podcasts with zero studio time. Feed it your typed writing and it will read it back to you in studio perfection. No ums and ahs.

Pair voice creation with facial mapping and voila.

Fake Tom Cruise.

Obama doesn’t swear like this, at least not in public.

Want to see Obama curse a blue streak? Here you go. It’s a one minute clip that will blow your mind a little. Part way through, you’ll get to see the guy who is actually making the fake. Using Obama’s face.

Can you see what some people would do with this kind of technology?

Some people will have fun with it. Hahaha joke videos.

But others? You know where I’m going, right?

How good is your imagination?

Know what Gal Gadot, Emma Watson, Natalie Portman, and Scarlett Johansson have in common besides being actresses?

They’ve all been used to create non-consensual porn videos. Deep fake porn. Of them. Without their permission. Yes, they look as real as the Tom Cruise videos. Except porn. Of them. Without their consent.

It happens more than you’d realize.

Sensity AI, a research company that has tracked online deepfake videos since 2018, has consistently found that between 90% and 95% are nonconsensual porn. About 90% of that is nonconsensual porn of women.

It’s going to get worse.

As fake Obama said so eloquently — we need to stay vigilant.

Now think of how deep fakes in politics. People will make deep fake videos of politicians they want to discredit. Saying things they’d never say. Except it’s not them. It’s fake. They’re probably already out there.

Think of the worse political moments you can. Then imagine politicians blaming them on deep fake. No, I didn’t say you could grab women by the p***y. That wasn’t me. I swear. It was fake. I wouldn’t say that.

And their followers will believe it. Because of course they will. No one will know what’s real and what’s fake.

Sure, technology companies are working on software to detect deep fakes. There’s software out there to detect ChatGPT, too. It fails a lot. How can we trust that Deep Fake detection will be any better?

Knowing who to believe is going to become a real challenge.

Knowing who to trust is going to be even harder.

Like fake Obama said, how we proceed is going to make the difference between whether we survive or create some f***ed up dystopia.

What happens when you can’t trust your eyes anymore? It’s a good question, you think?

I’ll leave you with a one min. clip of Morgan Freeman. It’s not really Morgan Freeman. It’s a deep fake. The end will boggle your mind.

As one viewer said on youtube, “Just when you thought nuclear weapons were the most dangerous things humans could create…

He’s not wrong. We didn’t think of the repercussion before we created the monster and now we can’t undo what we did.

And I don’t know what to do with that.

This is the way the world ends This is the way the world ends This is the way the world ends Not with a bang but a whimper. T.S. Eliot, The Hollow Men

AI
Technology
Tech
Videos
Movies
Recommended from ReadMedium