How do deepfakes create realistic fake videos and audio?

Deepfakes use tricks to make fake videos and audio look and sound real, just like when you put on a fun costume and pretend to be someone else.

Imagine your favorite cartoon character, they move their mouth when they talk. Deepfakes work in a similar way, but with real people. They take lots of pictures or videos of someone’s face and use special computer tools to figure out how their face moves when they speak. Then, the computer can make it look like that person is saying something new, even if they never said it before!

How It Works Like a Puzzle

Think of deepfakes as a puzzle with two parts:

  1. The face part, the computer learns what your face looks like when you smile, frown, or say “hello.”
  2. The voice part, the computer listens to how your voice sounds and copies it.

Then, the computer puts those pieces together to make a new video where someone’s face moves like they’re talking, even if they didn’t say those words!

It's like having a friend who can copy your voice and face so well that you wouldn't know if they were really you or just pretending. Pretty cool, right?

Take the quiz →

Examples

  1. A person's face is replaced with another in a video, making it look like they're saying something they didn't.
  2. Fake audio sounds like someone’s real voice, even though it was made by a computer.
  3. A celebrity appears in a video talking about a product they never promoted.

Ask a question

See also

Discussion

Recent activity