Back to identity glossary

Deepfake

A deepfake is any image, video, or audio that shows a person doing something they never actually did. The name is a portmanteau of the word “deep learning,” which is the machine-learning technique used to generate the images and the word “fake.” Deepfakes are also broadly called synthetic media

Fraudsters use deepfakes to bypass verification and authentication systems so they can:

  • Create fake accounts

  • Access accounts that aren’t theirs 

  • Access sensitive data (personal, financial, health, etc.)

  • Complete or authorize fraudulent transactions

  • Create a synthetic ID

  • and more

Frequently asked questions

To create a deepfake, a bad actor first feeds image or audio samples of a real person into a neural network or generative adversarial network (GAN). These samples are used to train the network, which can then generate synthetic media.

Although deepfakes are becoming harder to detect, some indicators to watch out for include:

  • Odd reflections

  • Irregular lighting

  • Flickering edges

  • Inconsistent color or shading

  • Small details being out of place

  • Audio that’s out of sync