Deepfake
AI-generated video, image or audio faking someone saying or doing what they never did.
A deepfake is synthetic video, image or audio generated by artificial intelligence to make someone appear to say or do something they never did. The technology has moved from novelty to fraud tool: convincing fake video of a public figure endorsing an investment, a manipulated clip of a chief executive authorising an urgent payment, or fabricated images used to pressure or blackmail.
Paired with voice cloning, a deepfake can make an impersonation feel unarguable, because we are wired to believe our own eyes and ears. Quality varies, and tells still exist — odd blinking, edges that shimmer, lip-sync that drifts, lighting that does not match — but these are fading fast and should not be relied on alone. The durable defence is procedural rather than visual: verify any surprising request through a separate, trusted channel, and never let a face or a voice on a screen override a check you would otherwise make. Treat urgency plus a request to move money as the warning, whatever the medium. Our article on deepfake and voice-cloning scams shows how these are used and how to stay a step ahead.