How do deepfakes work and what are their societal risks?

Deepfakes are like pretend faces on videos that can say and do things they never actually said or did.

Imagine you have a favorite cartoon character, let's say it's a silly dog who always says "Woof!" You know, the kind of dog who could be your friend if only he spoke English. Now imagine someone took pictures of that dog from different angles and made a video where he seems to talk to you in real time. That’s what deepfakes do, they take lots of photos or videos of a person and use them to make a new video where the person says things they never said before.

Think of it like drawing with markers. First, someone takes hundreds of photos of a person's face from different angles, kind of like how you might draw your friend on paper, one eye at a time, then the mouth, and so on. Then, using a special computer program, they put all those pieces together to make it look like the person is speaking or acting in a new video.

Sometimes people use deepfakes to trick others, maybe to make it seem like someone famous said something silly or even did something bad. That can cause confusion and fake news, which makes it hard for everyone to know what’s really true. It's like if your friend told you a secret, but then later on, you found out that they were just pretending!