AI
Deepfake
Last updated
Synthetic media (video, audio or images) of a real person, generated by AI without their consent, typically to deceive viewers.
Full definition
Deepfake refers to AI-generated synthetic media depicting a real person doing or saying something they did not actually do or say, created without their consent for deceptive purposes. The term combines 'deep learning' and 'fake'. Distinct from legitimate AI avatar use: deepfakes are non-consensual; consent-verified AI avatars and voice clones (like ElevenLabs Professional Voice Clone or HeyGen custom avatars) are not deepfakes. Major platforms actively prevent deepfake creation through identity verification. US states have passed deepfake legislation (CA, TX, VA, IL among others) covering non-consensual intimate imagery and political deepfakes. EU AI Act (effective 2026) requires labeling of AI-generated media.
Examples
- ·A political deepfake video showing a candidate saying something they never said — illegal in many US states.
- ·A consensual AI clone of a creator's own voice and face for content production — not a deepfake.
- ·Identity verification flows on ElevenLabs and HeyGen that prevent users from cloning public figures without consent.