Google just gave Google Vids the ability to put you on camera without you actually being on camera. Upload a selfie and a voice recording, and the tool builds a custom digital avatar that looks and sounds like you. According to TechCrunch AI, the update landed Thursday, and it arrives with a second headline change: Gemini Omni, Google’s multi-modal model, is now wired into Vids.
The timing says a lot. OpenAI’s Sora has shut down, and Google clearly thinks the appetite for starring in your own AI videos didn’t leave with it.
What actually shipped
- Personal avatars built from a selfie and a voice clip. No studio, no green screen, no recording session. You hand Vids two inputs and it generates a version of you that can present on your behalf. This is the feature that changes what Vids is for.
- Gemini Omni inside the editor. You write a prompt, upload reference images, and Omni mixes both into a finished video. The reference images matter here. Text-only prompting gives you whatever the model imagines. Feeding it your own visuals pulls the output toward something you can actually use at work.
- Fixes for footage you already shot. TechCrunch AI notes Omni can swap out a background, correct lighting on video recorded with your phone, or layer in effects. That’s the unglamorous stuff that eats hours in a normal edit.
- Step-by-step edits. You can adjust a video as you go instead of regenerating the whole thing from scratch. Anyone who has burned an afternoon re-rolling a generation to fix one detail understands why this is the quietly important item on the list.
Vids isn’t a slideshow tool anymore
Vids started life as an AI-assisted workplace presentation tool. Slide-adjacent. Safe. These updates push it toward an all-in-one video creation platform, and TechCrunch AI frames the shift the same way.
Google is still signaling the business use case by keeping Vids inside Google Workspace. Company updates. Training videos. The internal comms nobody wants to film twice. But personal avatars and conversational editing drag it into a different neighborhood entirely.
The competitive read
That neighborhood already has residents. TechCrunch AI names HeyGen, Synthesia, Captions, and D-ID as the tools Vids now bumps against.
What stands out here is distribution, not features. Those startups built their businesses on avatar generation being a standalone product you go out and buy. Google just made a version of it a checkbox inside a suite that millions of companies already pay for. You don’t need procurement approval for something already sitting in your Workspace tab. That’s a hard problem for a point solution to solve with better output quality alone.
The guardrails
Google put real fences around this one:
- Avatars are tied to the account holder’s likeness and locked to their Google account
- Every output carries an invisible SynthID watermark
- Access to personal avatars is limited to users 18 and older, in certain regions only
TechCrunch AI made the obvious joke, and it lands: nobody’s going to be generating bizarre AI videos of Sundar Pichai the way Sam Altman let people do with Sora. Google watched that experiment run and drew a different conclusion.
Why it matters
The identity lock is the actual story. Sora’s open-season approach to likeness generated enormous attention and enormous liability, and it’s gone now. Google is shipping the same core capability with consent baked into the architecture rather than bolted on as a moderation policy. You can only be yourself.
That’s a slower, more boring product. It’s also the one an enterprise legal team will approve.
The open questions are the ones Google didn’t answer: which regions, how convincing the avatars actually look in practice, and whether “tied to your Google account” holds up once people start looking for ways around it. Watch the avatar quality bar. If it clears the uncanny valley for standard corporate video, a lot of paid avatar subscriptions start looking optional.
Full details are in the original TechCrunch AI report.