Easy, undetectable, and terrifying. What will happen now that anyone can create a completely realistic fake video of anyone saying anything?


(photo: geralt / 19928 bilder)

Deepfakes are here, and boy are they scary.

A seamlessly edited fake video posted recently to social-media giant, Instagram showed Facebook founder Mark Zuckerberg giving a chillingly sinister speech about the unlimited power of Facebook.

“Imagine this for a second: One man, with total control of billions of people’s stolen data, all their secrets, their lives, their futures. I owe it all to Spectre. Spectre showed me that whoever controls the data, controls the future.” — Deepfake Mark Zuckerberg, on the faked video.

The video was created using a real video of Zuckerberg making a statement to CNN in 2017 after news broke of Russian interference with the 2016 election. A company called CannyAI created the software used for the faked video. The software uses new video dialogue replacement technology.

Deepfake videos can be made from even a single image of a face.

This particular video is only truly believable if you don’t listen to the volume: The voice is clearly not Zuckerberg’s. But soon, this barrier too will fall and complete video realism, and all its implications are upon us.

Facebook has previously promised not to remove such videos.

“We will treat this content the same way we treat all misinformation on Instagram. If third-party fact-checkers mark it as false, we will filter it from Instagram’s recommendation surfaces like Explore and hashtag pages.” — a spokesperson for Instagram told Motherboard

When deepfake videos are indistinguishable from the real thing, will the tech giants still be able to maintain that policy?

Perhaps more important than the moral implications of such a strategy, are the practical considerations:

How does any company moderate over 2 billion people?

Recently, a doctored video of Speaker of the House Nancy Pelosi began to circulate on the internet. In a move considered by some to be questionable, one person responsible for the perpetuating the video and distributing it widely over social media channels as genuine was ‘doxxed’ by an investigative journalist determined to find the source of the edited video.

The social media user who shared the Pelosi video, had his identity, along with his address and personal information published by by the press in connection to the post.

This Person Does Not Exist

The next generation of advanced AI can already make realistic human faces.

Websites like www.thispersondoesnotexist.com can generate endless human faces that are indistinguishable from real human faces. Soon, next generations of these same algorithms will be able to generate live action movies with completely realistic human actors.

What will the world do with technology that can make a video of anyone doing anything? The possibilities are as endless as they are eye-popping. As comedian Dave Chapelle once quipped, “If the internet was a real place, it would be way to disgusting to go there.”

Officials and the judiciary will soon need to deal with a plethora of new and novel court cases in which old laws, and old institutions guided by an even older document, will need to be applied. And carefully.

When and if these deepfake videos result in real-world consequences for the slandered party including loss of livelihood, property damage, even loss of life, the courts and lawmakers will need to respond effectively.

Eventually, the deepfake video of Zuckerberg boasting about the unlimited power of his company was removed after a request from CNN. Because the video was so realistic, and because fake-zuckerberg was presented as if he were giving a very-realistic looking news briefing, CNN’s logo was figured prominently.

The video was removed, not for spreading misinformation that could be potentially damaging to a U.S. citizen, in this case Zuckerberg, or a U.S. company, in this case Facebook, but for property rights infringement.

The video was removed under existing copyright laws.

It is a somewhat novel approach to the problem. But video fakers could avoid reproducing logos for future videos and avoid this potential trap.

The longer a video stays available on the internet, the wider it is spread. But once a deepfake video is loaded onto the internet, its contents, no matter how deranged, damaging, or discrediting, will live forever.

How will public figures, politicians like Nancy Pelosi who trade on their reputations protect themselves from embarrassing misinformation that could get them arrested, defamed, or worse, defeated in an election.

How far does the right to free speech go in the Information Age?

Some cases are already testing the limits.

The California man whose Wichita Kansas swatting prank inadvertently cost a man his life recently received a hefty jail sentence. Though the defendant was more than a thousand miles away from the crime, he was still found guilty. The swatter did nothing more that tell authorities something that wasn’t true via online message- that a man at the address he gave was holding hostages at gunpoint and had already killed someone- the jury still found him culpable in the man’s death.

Companies like Facebook and Instagram will follow the letter of the law. That means, it is up to legislators and the tech industry to respond with laws designed to limit, and mitigate, the damage that might be caused by deepfake videos.

In a changing world, internet users are a new classification of people who must have their rights and property protected under the law. Even property as ephemeral as someone’s reputation deserves protection, which is why we have laws against slander and libel.

No one is safe from deepfakes and deepfake videos will need their own classification. Lawmakers delay at their peril, as they are likely to be the first targeted with embarrassing attacks.

The face you save may be your own.

(contributing writer, Brooke Bell)