YouTube Shorts Adds AI Avatars That Replicate Your Voice and Appearance

YouTube is now rolling out an AI avatar feature that enables users to create photorealistic digital versions of themselves for use in Shorts videos, with full voice synthesis capabilities.

How It Works

Users create their avatar through a one-time setup process available in the YouTube app and YouTube Create. The process requires capturing a "live selfie" by recording your face and voice while reading several prompts. This generates a photorealistic avatar that can be placed in YouTube Shorts.

Each prompt-based generation creates clips up to 8 seconds long. Users can create multiple clips back-to-back and retake the setup at any time to update their appearance. The initial setup is required only once.

Features and Limitations

The avatar generation is limited to prompt-based creation—users cannot use uploaded images or existing video footage. Only the avatar's owner can create videos with it; no other user can generate content using someone else's avatar.

All avatar videos will include watermarks and digital labels (SynthID and C2PA) along with AI-generated content disclosures. YouTube states the feature "gives users an easier way to include themselves safely and securely in videos."

Users can delete their avatars at any time. YouTube will automatically delete avatars after 3 years of inactivity. However, existing Shorts featuring the avatar will remain live until the actual clip is manually deleted.

Availability and Access

The feature is rolling out globally today (outside Europe) for users 18 and older. Full availability is expected within days. To access it, users open the Create '+' button in the YouTube app, tap the Gemini spark icon, select "Create video," and choose "Make a video with my avatar." The feature is also accessible via Remix > Reimagine > Add me to this scene.

You must be the owner of an existing YouTube channel to create an avatar.

Technical Foundation

This feature builds on Google's Veo video generation models already integrated into YouTube Shorts. YouTube previously offered ingredients-to-video capabilities requiring image uploads; voice synthesis represents the new addition here.

What This Means

YouTube is integrating voice and video synthesis directly into Shorts creation, lowering the barrier for users who want to appear in videos without recording themselves. The inclusion of disclosure labels and non-transferability protections suggests Google is attempting to address deepfake concerns upfront. However, the 8-second limitation and avatar-only generation (no remix of existing footage) significantly constrains use cases compared to full video generation models.