YouTube Shorts AI Creates Personalized Avatars
▼ Summary
– YouTube has launched a feature allowing users 18+ globally (excluding Europe) to create a photorealistic AI avatar of themselves for use in Shorts.
– The avatar is created by recording a live selfie and voice sample in the app, a one-time setup that generates prompt-based clips up to 8 seconds long.
– This is a new application of Google’s Veo models, distinct for incorporating a user’s own voice, unlike previous image-based video tools.
– For safety, the avatar is locked to the creator’s account, all outputs have AI disclosure labels, and the source data is deleted after three years of inactivity.
– The feature is accessed via the Create button in the YouTube app and requires users to own an existing YouTube channel.
YouTube is introducing a new tool that allows creators to generate a personalized AI avatar for their Shorts content. This feature, which was previewed earlier in 2024, enables users to produce a digital likeness that mirrors their own appearance and voice. The process is designed to offer a secure and simplified method for creators to appear in videos without filming themselves directly.
The setup is accessible through the primary YouTube app and the YouTube Create platform. To begin, users record a brief “live selfie” video, capturing their facial features and voice by reading a few on-screen lines. This one-time session produces a photorealistic avatar that can be inserted into YouTube Shorts. Each AI-generated clip can last up to eight seconds, and creators can string multiple segments together seamlessly. The initial model can be updated at any time by re-recording the selfie.
This innovation builds upon Google’s existing Veo AI models within the Shorts ecosystem. While a previous feature allowed for video creation from a static photo, the new capability integrates a synthesized voice, marking a significant advancement. YouTube emphasizes that the tool provides a safer way for individuals to participate in video content, as the underlying selfie data is used exclusively for avatar generation and cannot be repurposed by others.
From a safety and security standpoint, YouTube has implemented several safeguards. The original selfie video and audio are processed solely for creating the user’s unique avatar. No other account can utilize this digital likeness to produce original Shorts. Users retain control, with the option to delete their avatar data at any time; inactive avatars are automatically removed after three years. However, any previously published Shorts featuring the avatar will remain online until the creator manually deletes those specific videos.
To ensure transparency, all content generated with this tool will carry visible AI disclosure labels. This includes digital watermarks and industry-standard identifiers like SynthID and C2PA, clearly signaling that the video features synthetic media.
For users with access, the feature is found within the creation workflow. Tapping the ‘+’ Create button in the YouTube app reveals a Gemini sparkle icon. Selecting “Create video” and then locating the “Make a video with my avatar” option initiates the process. It is also available through the Remix menu under the “Reimagine” function, which includes an “Add me to this scene” choice.
The global rollout for this feature begins immediately for users aged 18 and over, though it is not initially available in Europe. Full accessibility is expected in the coming days. A prerequisite for using the tool is ownership of an existing YouTube channel.
(Source: 9to5google.com)
