Overview
- YouTube’s avatar feature, which began rolling out Wednesday, is reaching mobile users outside Europe who are 18 or older and own a channel, with wider access expected in the coming days.
- Creators record a live selfie and read short prompts to capture face and voice, then type a prompt to generate about eight seconds of video that can be chained into longer Shorts.
- Only the account holder can use their avatar, selfie and voice data are used only to build that avatar, and creators can delete or retake it at any time.
- YouTube labels every avatar clip as AI-generated with visible watermarks and provenance tags like SynthID and C2PA, though outlets note ongoing worries about deepfakes and misuse despite these guardrails.
- Reporting points to Google’s Veo 3.1 video model and Gemini-powered tools behind the feature, which extends YouTube’s recent AI additions as rivals reassess costly, controversial video generation.