what is a deepfake?

An AI-generated image, audio, or video that resembles a real person or event and can be mistaken for genuine. What the law requires, and where the line is drawn against ordinary AI images.

A deepfake is an AI-generated image, audio, or video that depicts a real person or event so convincingly it can be mistaken for genuine: a politician saying something they never said, a voice on the phone that sounds like the finance director’s, a photo of an event that never happened.

Since 2 August 2026, the EU AI Act requires deepfakes to be visibly labelled: anyone seeing or hearing the material must understand it has been manipulated. The labelling obligation lies with whoever publishes it, not just whoever created it. The line between deepfakes and ordinary AI images is the risk of confusion: a generated illustration of an office is not a deepfake, but a generated photo of a named real person is.

The practical protection has two parts. One is labelling: images EuroWork generates carry a machine-readable note about their origin in the file itself, and downloads include a ready-made visible label. The other is vigilance in the other direction: when a voice on the phone asks for a quick payment, the sensible follow-up question is no longer "who is this?" but "can I call you back on the number I have?".

EuroWork

Your employees are already using AI, often in private accounts where no one sees what gets pasted in. EuroWork gives the same AI help in one place where you are in control: sensitive data is caught before it is sent, processing stays in the EU, every answer shows its price and management sets the rules.

See how EuroWork works →