What is AI avatar?

An AI avatar is a synthetic on-screen presenter — a face and voice generated or animated by a model — used to deliver scripted video without filming a person. Most implementations fall into two groups. Image-to-video systems animate a still photograph, driving lip movement and limited head motion from an audio track.

Fully synthetic avatars generate the presenter from scratch, offering more control over appearance and motion at the cost of a look that viewers more readily identify as artificial. The commercial appeal is that a script change no longer requires a reshoot, which makes localisation, personalisation, and high-variant testing tractable in a way filmed presenters never were. The constraints are worth naming. Realism degrades over longer takes as gesture repetition becomes noticeable, so short-form performs better than long. Consent and likeness rights apply whenever the source is a real person's face, and using someone's likeness without a licence creates legal exposure regardless of how the video was produced. Platform labelling policies for synthetic media apply to this format specifically, since it depicts a person who appears to be speaking.