The sinister rise of deepfake videos

New frontiers: Twitter has pledged to remove deepfake content and block its publishers.

Will deepfakes destroy democracy? In a viral video, Mark Zuckerberg lays out his scheme for world domination using stolen data. But the footage isn’t real — it was created with AI technology.

“Imagine this for a second: one man with total control of billions of people’s stolen data; all their secrets, their lives, their futures.”

Mark Zuckerberg, founder of Facebook, stares into the camera.

“I owe it all to Spectre. Spectre showed me that whoever controls the data, controls the future.”

It’s a sinister message from the owner of a social network used by 2.4 billion people. But this video is not real. It’s a deepfake.

Deepfakes are produced with a technology called “deep learning”, which uses neural networks that mimic the human brain. Artificial intelligence (AI) programmes can be fed images of a person, and can then produce footage of almost anyone saying or doing anything.

The video of Zuckerberg is part of an installation called Spectre, created by artists Bill Posters and Daniel Howe. They want to draw attention to Facebook’s policy on deepfakes. The social networking giant said that the video did not violate any rules and would not be removed.

It came just weeks after viral footage appeared to show US Democrat politician Nancy Pelosi slurring. That video was made using a more rudimentary means, called a “cheapfake”.

But sophisticated technology means that deepfakes are getting better and harder to spot.

“They take the threat of fake news even higher, as seemingly anyone can now have the ability to literally and convincingly put words in someone else’s mouth,” says Gil Becker, CEO of content platform Anyclip.

According to the World Economic Forum, deepfakes are increasingly going to “challenge our trust in reality”.

They could be used to discredit honest politicians, or give leaders an excuse to dismiss reality as fake news.

As deepfakes spread through social media, policymakers are searching for ways to authenticate videos. In the US, a proposed law would force the creators of fake videos to verify their content with a digital watermark.

But the plan has quickly been dismissed as ineffective, with many fearing that we are powerless to stop the spread of disinformation.

Broken reality?

Will deepfakes destroy democracy? They herald a future in which we can no longer trust our own eyes and ears — a future in which truth and reality are subjective, used by different tribes again each other. Debate and democracy are based on a shared understanding of the truth, even if our approaches differ. The very basis of our society is under threat.

Or are these fears hysterical? Doctored footage is nothing new. In the world of fake news, we already know to be sceptical of what we see online. They could even offer us a wake-up call to be more critical, and encourage the media to come up with innovative technological solutions to restore trust and verify the truth.

You Decide

  1. Are deepfakes a threat to democracy?
  2. Could deepfake technology have any benefits?

Activities

  1. Make a list of reliable news websites. Are there any you and your classmates disagree about?
  2. Make a leaflet warning people about the rise of deepfakes, explaining how they are made and why they are a threat.

Some People Say...

“The truth is incontrovertible. Malice may attack it, ignorance may deride it — but, in the end, there it is.”

Winston Churchill (1874-1965)

What do you think?

Q & A

What do we know?
Deepfakes are videos that have been manipulated to show the subject saying or doing things that they never did. Artists Bill Posters and Daniel Howe created a deepfake video of Facebook founder Mark Zuckerberg for an installation called Spectre, which drew attention to Facebook’s policy of not removing fake videos from its platform.
What do we not know?
How prevalent deepfakes will become, and what the consequences will be. The technology has emerged rapidly over the past two years. Some AI programmes can detect and discredit deepfake videos, but they are becoming increasingly hard to spot.

Word Watch

Neural networks
These AI systems mimic the way that neurons in the human brain connect, which allows computers to “learn” like humans.
Nancy Pelosi
Speaker of the US House of Representatives, which is one of the two chambers in Congress.
Watermark
A marker embedded into a digital file that could mark it out as verified. Media organisations may start using digital watermarks to show that their content is truthful.
Disinformation
The deliberate, malicious spread of false information. When someone spreads or shares falsehoods without realising, it is called misinformation.

Subjects

PDF Download

Please click on "Print view" at the top of the page to see a print friendly version of the article.