Decoder. plain-English AI glossary

Deepfake

● Core

Also called Deepfake

A convincingly fake image, video, or voice generated by AI to show someone doing or saying something they never did.

Think of it like

A flawless forged photograph — except now the forger works in minutes and the seams are nearly invisible.

Example

A cloned voice of a company’s CEO calls the finance team and authorizes an urgent wire transfer that never should have happened.

How it actually works

Generative models can now synthesize faces, voices, and video realistic enough to fool people at a glance. Deepfakes power fraud, non-consensual imagery, and political disinformation, and detection is a losing arms race — as detectors improve, so do the fakes. The more durable defenses lean on provenance (watermarking, signed authentic media) and on plain skepticism about unverified footage, rather than on spotting the fake after the fact.

For product teams

A trust-eroding threat that raises the bar on verifying anyone’s identity or any piece of media.

For engineers

Realistic synthetic audio/video; post-hoc detection is an arms race, so provenance (watermarks, signed capture) is the more durable defense.

Related

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome