meta_pixel
Tapesearch Logo
Log in
Today, Explained

The Deep Fake

Today, Explained

Vox

Politics, Daily News, News

4.3 • 10.3K Ratings

🗓️ 1 March 2018

⏱️ 19 minutes

🧾️ Download transcript

Summary

There’s a new kind of algorithm that allows you to take a video of one person and map the face of another person onto his or her body. Not surprisingly, it’s being used to map celebrities’ faces onto the bodies of porn stars having sex. Vox’s Aja Romano tells Sean Rameswaram how “deepfakes” are spreading across the internet. Plus computer scientist Peter Eckersley explores how the same technology could tear our society apart in bigger ways. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Transcript

Click on a timestamp to play from that location

0:00.0

Hey, happy March! You've heard of March Madness, how about a March mattress? The weather's

0:05.1

turning, your bed is yearning? Check out Matches Firm. They're open all month, except at

0:10.3

nighttime because they're big on sleep. But their website never closes at all. Head to

0:14.8

mattressfirm.com slash podcast to learn how you can improve your sleep.

0:25.5

This is today explained, I'm Sean Ramos from.

0:38.7

There's this crazy thing happening online right now that could make you question whether

0:43.2

what you're hearing, whether what you're seeing is real. It could affect who you trust,

0:49.5

it could affect elections, but right now it's mostly affecting porn. It kind of goes

0:55.3

back to this long standing rule of the internet. If it exists, there's porn of it. Now we know

1:00.8

that even if it doesn't exist, there's porn of it. This is Asia Romano. I am an internet

1:06.2

culture reporter for Vox. Asia's been writing about algorithms lately. The ability to build your

1:10.8

own predictive algorithm has been there for years. Predictive algorithms take some stuff and

1:16.2

with that stuff, they can imagine totally new stuff. Recently, people have been using these

1:22.0

predictive algorithms to make fake videos of real people. It's basically learning what

1:27.5

your face looks like and then predicting what your face would look like mapped onto another

1:32.9

image that is doing something else. Like if I were to show the computer the neural network

1:36.7

enough photos of your face, it would learn what you look like based on all of the photos

1:42.1

that I give it and then predict what your face would look like doing something else that

1:47.4

I feed it. So if I give it a video of, I don't know, someone chopping wood, it would

1:51.6

then be able to say this is what he would look like chopping wood. Does that make sense?

1:55.9

No, because why would anyone want to watch a fake video of a podcast host chopping wood?

2:01.7

But because there are a lot of perviedudes on the internet, take one guess what they've

...

Please login to see the full transcript.

Disclaimer: The podcast and artwork embedded on this page are from Vox, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of Vox and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.

Copyright © Tapesearch 2026.