Google Gemini adds an avatar tool that makes AI videos of you

In short: Google’s Gemini app now lets paying users make short AI videos featuring a digital copy of themselves.

What happened

A WIRED writer tested a new Gemini feature called “avatars,” which creates a digital version of you that can appear in AI-made videos. Think of it like a reusable character model in a video game, except it is built from your face and voice.

To set it up, the writer used a phone camera in a well-lit room, read a series of two-digit numbers, and slowly turned their head side to side. The setup took about five minutes. After that, Gemini could generate two 10-second clips showing the writer doing things they never did in real life, like singing “Happy Birthday” to a dinosaur in San Francisco’s Dolores Park and surfing under the Golden Gate Bridge.

The writer said the results were both impressive and unsettling. The videos had errors, like odd clothing choices and some jumbled motion, but the face and voice looked close enough to feel “unnervingly” real. The feature is powered by Google’s “Omni” video model, which is the system Gemini uses to create these videos, and it is only available to subscribers. The writer paid $20 per month for Google’s AI Pro plan and hit usage limits after a few prompts, with limits resetting every five hours.

Google told WIRED it only allows adult users to create videos using their own avatar. A Google DeepMind product lead said the company tries to prevent harm while not blocking harmless uses.

Why it matters

Tools like this can make it easier for regular people to create realistic videos, but they also raise fears about deepfakes, meaning fake videos that look real. Even with limits, a lifelike “you” that can be placed anywhere could change how people think about proof, trust, and consent online.

Source: Wired