When people hear "AI," what often comes to mind?

Most likely, they think of applications like ChatGPT, which can frequently pass AI detection tests that many humans struggle with, and that's not an exaggeration.

Alternatively, they might consider AI art generators that allow users to create artwork with just a few prompts. Imagine painting like Picasso with minimal effort!

However, what tends to be overlooked in discussions about AI?

Video content, particularly deepfake videos.

And rightly so, as they are alarmingly realistic.

Let’s clarify:

This is not the real Tom Cruise.

It’s a deepfake representation sourced from the TikTok account known as #deeptomcruise.

Can you even grasp this?

I must emphasize: This is a fabrication. It is not Tom Cruise.

The audience is left in disbelief.

From the intricate details like the creases on his forehead when he raises his brows to the way he laughs and adjusts his hair, it all appears to be Tom Cruise.

Yet, it’s not.

Viewers are left saying that if this isn't the real Tom Cruise, they will be shocked. Many refuse to accept it; they genuinely believe it’s him, convinced that someone is trying to deceive them.

Surely, it must be Tom Cruise, right?

Wrong. It’s not him.

If you check the date on the comments, you’ll see this was two years ago.

Two years.

How much do you think deepfake technology has advanced since the inception of the deeptomcruise account?

Here’s another astonishing clip with 28 million views.

Does Tom Cruise speak Spanish? Who knows! But fake Tom Cruise does.

He converses in Spanish, chuckles at himself, and seamlessly switches between English and Spanish with such realism that it's hard to believe.

One viewer insists they "know" it's the actual Tom Cruise because his teeth are slightly misaligned, like Cruise's. For them, that's the key to its authenticity, assuming AI can't replicate such nuances.

But it can.

That’s precisely how deepfakes operate.

The account boasts 5.2 million followers. Regardless of the clip, people continuously question its authenticity. But remember, it’s fake.

Deepfakes are incredibly convincing.

In the last video, he stumbles, chuckles, and shares a joke...

Again, the comments reveal astonishment.

"Disturbingly accurate. Is this truly Tom Cruise? Insane. It’s a bit frightening how realistic it appears." [Comments from YouTube, source]

These are the kinds of reactions people have, and rightfully so. The resemblance is uncanny.

So, how does this work?

Essentially, it involves facial mapping. All that’s needed is a multitude of photographs of the individual to be replicated. The more images, the better.

Here’s the process: They collect numerous photos of Tom Cruise and input them into an AI machine learning program, similar to how ChatGPT was trained on internet text.

And, of course, there is no shortage of Tom Cruise photos.

The AI analyzes these images to learn how his face moves, what expressions he makes, and how his eyebrows react when he smiles. It "learns" his facial expressions just like ChatGPT learns word combinations.

Next, they require a model. This person need not resemble Tom Cruise in any way. The technology then maps Tom Cruise's face onto this model's face.

And voilà — deepfake!

The second part involves AI voice replication.

Did you know that’s possible?

There are AI tools as user-friendly as ChatGPT. You create an account and generate an AI version of your own voice.

Each program operates slightly differently.

The most accurate ones require users to read numerous sentences—around 70 or 100—to teach the AI how to mimic their speech patterns.

You can create podcasts without needing studio time. Just input your text, and it reads it back to you with studio-quality delivery, no filler words.

Combine voice generation with facial mapping, and you get... fake Tom Cruise.

What about Obama? He doesn’t typically use strong language in public.

Want to see Obama use some colorful language? Here’s a one-minute clip that might shock you. Halfway through, you’ll see the individual behind the deepfake, using Obama’s likeness.

Can you imagine how some might exploit this technology?

Some will create amusing parody videos.

But others? You likely see where this is headed.

Consider this: Gal Gadot, Emma Watson, Natalie Portman, and Scarlett Johansson have all been victims of non-consensual deepfake pornography.

These videos appear as realistic as the Tom Cruise clips, but they portray them in objectionable contexts without their consent.

This issue is more widespread than you might think.

Sensity AI, a research organization tracking online deepfake content since 2018, consistently reports that 90% to 95% of such videos feature non-consensual pornography, with about 90% targeting women.

The situation will likely worsen.

As the fake Obama wisely noted, we need to remain vigilant.

Consider the implications of deepfakes in politics. Individuals could fabricate videos of politicians making statements they would never utter.

But it’s not really them; it’s a deception. Such videos are probably already circulating.

Think of the most scandalous political moments you can imagine. Now envision politicians attributing these to deepfakes. "No, I never made that statement; it wasn’t me. It was fake."

And their supporters will likely buy into it.

Determining what’s real and what’s fabricated will become increasingly challenging.

While tech companies are developing software to identify deepfakes, similar tools exist for detecting ChatGPT output, and they often fail. Can we rely on deepfake detection to be any more effective?

Knowing whom to trust will become a significant dilemma.

As fake Obama articulated, our approach will determine whether we navigate to safety or descend into a dystopian reality.

What occurs when you can no longer trust what you see?

That’s a thought-provoking question, don’t you think?

I’ll conclude with a one-minute video featuring Morgan Freeman.

It’s not really him; it’s a deepfake. The ending will leave you astonished.

As one viewer on YouTube remarked, “Just when you thought nuclear weapons were the most perilous creations of humanity…”

They are not incorrect. We didn’t anticipate the consequences of our innovations and now face the aftermath.

And honestly, I’m unsure how to address 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