Google Vids adds AI avatars and Gemini Omni to challenge video creation startups

Google has expanded Google Vids with two significant capabilities: custom AI avatars based on user likenesses and integration of its Gemini Omni multimodal AI model for video creation and editing.

The avatars, announced Thursday, are generated from a selfie and voice recording. They're tied to the account holder's identity, watermarked with Google's SynthID technology, and restricted to users 18 and older in select regions. The company frames this as a way to create personalized video content while preventing misuse—a response to concerns that emerged around deepfakes and synthetic media.

Gemini Omni integration lets users create videos by combining written prompts with reference images. The model can also modify existing videos: adjusting backgrounds, fixing lighting in phone recordings, and adding effects. A step-by-step editing workflow means users can make iterative changes without restarting from scratch.

These updates signal Google's ambitions to move Vids beyond its original positioning as an AI-assisted workplace presentation tool. By embedding it in Google Workspace, the company targets business use cases like company updates and training videos. But the personalized avatars and conversational editing features put Vids in direct competition with established players like HeyGen, Synthesia, Captions, and D-ID.

The timing is notable: OpenAI's Sora video generation tool shut down earlier this year, leaving space for competitors to expand. Google's move suggests the market for AI video creation remains active despite that exit.