Google Adds Self-Clone Avatars to Flow as Omni Flash Replaces Veo for AI Video

3 articles · Updated · WIRED · May 19

Google unveiled a new Flow avatar feature at I/O, letting users scan themselves and insert AI versions of their face and voice into generated video clips.
Omni Flash now powers Flow in place of Veo, aiming to keep characters more consistent across edits and preserve avatar details when users change backgrounds, clothing or other prompts.
Users create an avatar by scanning a QR code, recording themselves saying a string of numbers and turning their head to capture multiple angles; all Omni-generated videos carry SynthID watermarks.
Google is also bringing avatars to Gemini and YouTube, but unlike OpenAI’s short-lived Sora app, the initial rollout is limited to generating AI versions of the user, not other people.
The launch fits Google’s broader push to make AI creation and automated workflows mainstream for creators, even as generative video tools face backlash from audiences who see them as inauthentic.

AI avatars can now replace actors in videos. Is this the future of creative freedom or the end of authentic human performance?

Google can now create your digital twin. But who truly owns and controls your AI avatar and all its personal data?

Google watermarks every AI video. In the escalating race against deepfakes, can any digital watermark truly remain unbreakable?