Fake Elon Musk wants to scam you.

In a widely shared “deepfake” video, the billionaire Tesla founder appeared to be promoting a new way to earn extra money in crypto.

It was a total scam. Musk had nothing to do with the video. But it fooled plenty of people. Musk would set the record straight on Twitter saying it was “def[initely] not me.”

Deepfake videos are fake but realistic-looking videos created using artificial intelligence technology. They can make it look like anyone is saying or doing anything… and they’re increasingly being used to fool people online.

If you find this disturbing… you’re not alone. From "catfishing” to Nigerian princes, the internet has always been used to scam certain people. But until recently, video was the one thing that couldn’t be easily falsified.

Thanks to rapidly advancing technology, that’s changed for good. And it’s opened up a new can of worms.

Could deepfakes fool you?

Could they swing an election?

Today, tech expert Chris Wood sheds light on deepfakes, including how investors should think about this trend…

***

Chris Reilly: Chris… what were your initial thoughts on the viral deepfake video of Elon Musk?

Chris Wood: The quality wasn’t even that good. If a low-quality deepfake like that can go viral and convince anyone that’s Elon Musk talking, that’s scary.

Because the good deepfakes produced today are much better and more convincing… and they’re only getting better and easier to produce every day.

Reilly: Can you shed some light on the tech behind deepfakes?

Wood: The term “deepfake” comes from a combination of the terms “deep learning” and “fake.”

Deep learning is a term from the artificial intelligence (AI) world. It’s a way to teach computers to learn like people do.

To make a deepfake video you “train” a computer with images and videos of a famous person and an actor. The computer learns how to seamlessly swap the actor’s face for the famous person’s face in a video. It also learns to perfectly match the actor’s facial expressions.

So, with deepfakes you can make anyone appear to say and do just about anything.

Last year a series of pretty convincing Tom Cruise deepfakes went viral on TikTok. Here’s a link to them.

Reilly: Where is this trend headed?

Wood: Unfortunately, deepfakes are growing exponentially.

A study I read last year reported the number of deepfake videos online doubled every six months from 2018 through 2020. And that pace has picked up since then.

In addition to those famous deepfake Tom Cruise videos, which racked up tens of millions of views, we’ve seen viral deepfakes showing Meta’s (formerly Facebook) founder and CEO Mark Zuckerberg giving a sinister speech about Facebook having “total control of billions of people’s stolen data.” And Barrack Obama calling Donald Trump a four-letter word.

Meanwhile, as the technology advances, these videos are going to spread like wildfire. Like most powerful technologies, it can be used for good, bad, and in between. For the bad guys out there, it’s too tempting not to weaponize.

Reilly: So far it seems like these videos are doing more harm than good… Are there any positives to deepfake videos?

Wood: Absolutely.

Think of education. We could create deepfakes of historical figures like Abraham Lincoln and animate their speeches to bring history to life for students.

This tech is also showing a ton of promise in the advertising and film industries.

In fact, the guys who made the Tom Cruise deepfakes launched a company called Metaphysic last year. It uses the same deepfake tech to make otherwise impossible ads and restore old films.

One of Metaphysic’s deepfake projects was a Gillette campaign that recreated a young NFL player Deion Sanders shaving with a Gillette razor on his draft day in 1989.

Metaphysic envisions using deepfakes to do everything from making older entertainers appear younger… to creating video doubles of famous people. These videos can be used to make commercials and other content—without them needing to be physically on set.

Reilly: What are the big-picture implications? Should readers be worried about deepfakes?

Could they, for example, swing an election?

Wood: Good question. Some analysts think that as deepfakes spread, most folks will lose all trust in digital media.

That would be a massive problem. But I think these analysts are shortsighted.

Folks must learn to think more critically about what they see online. And yes, it’s possible deepfakes could become a big problem.

But lots of solutions will come as well. That’s what always happens in tech.

Reilly: What sort of solutions are out there? Are there ways to invest in them?

Wood: Yes, certainly on the security side. Companies that create software able to verify photos and videos to bust deepfakes could be great investments.

A startup called Truepic developed a software platform to help do just that. Its tech can figure out what’s authentic in a digital photo or video and what (if anything) has been manipulated.

Truepic provides its tech to more than 100 customers already. Insurance companies, for example, use it to screen smartphone photos sent in from an accident.

Truepic is still private. But it’s received funding from some big, prominent investors including Microsoft’s venture capital arm.

More and more companies will be developing solutions like these, and surely some of those companies will go public. The space is too new for a public investor to put money to work today. But a lot of exciting opportunities are coming down the pike.

Chris Reilly
Executive Editor, RiskHedge