Google began rolling out a personal-avatar feature in Google Vids on July 16. You record your face and voice once, then Gemini Omni can generate new clips in which a digital version of you speaks and moves. For a solo creator who keeps postponing video because filming takes too long, that is a real production shortcut.
It also turns your likeness into a reusable business asset. A logo can be replaced. Your face and voice are harder to take back once clips have been downloaded, shared, or handed to a contractor. The sensible move is to decide how the avatar may be used before you make the first video.
How the feature works
Google announced a gradual rollout beginning July 16 for Rapid Release Workspace domains and August 5 for Scheduled Release domains. Its published eligibility list includes certain Business, Enterprise, Education, nonprofit, and Google AI Pro and Ultra plans. The launch terms limit the feature to adults using English and exclude the European Economic Area, Switzerland, and the United Kingdom. Check the current help page and your own account before assuming the feature is available to you.
Creating the avatar requires a phone or tablet. Google asks the account owner to record a face and voice sample and complete on-screen actions used for verification. After that, the avatar appears as an ingredient in Google Vids. You can give it a script, describe the setting and movement, choose a portrait or landscape format, and generate a clip.
The wider Omni tools in Vids can also use reference images, animate a still image, edit an uploaded video, and extend a generated scene. Google’s current help page lists generated clips at 720p and 24 frames per second. It says most users can generate up to 50 videos per month, although limits vary by plan and can change. This is built for short production work, not a one-click replacement for a finished video editor.
The part small businesses should decide now
Google says the personal-avatar feature is on by default for eligible Workspace domains, while administrators can disable it. If you run your own Workspace account, you may be both the user and the administrator. Do not wait for a setting to make the policy decision for you.
Start with ownership. Only the person whose face and voice are being recorded should complete the setup. Do not build an avatar for an employee, client, family member, or collaborator from files they sent for another purpose. If a client wants an avatar, put the approved uses, review process, and deletion plan in writing.
Then decide what viewers will be told. Google’s generative AI policy prohibits deceptive impersonation without explicit disclosure and prohibits falsely presenting generated content as entirely human-made in order to deceive. That is the minimum. For a practical trust rule, label an avatar clip whenever a reasonable viewer could believe you personally recorded that scene or said those exact words on camera. A short note such as ‘Created with my verified AI avatar; script reviewed by me’ is clear without becoming the headline.
Google has also expanded SynthID, its invisible watermarking system for AI-generated media. Technical provenance can help platforms and viewers investigate where a file came from. It does not replace a visible disclosure in a sales video, testimonial, course lesson, or public statement where the context matters.
A five-rule avatar policy you can use
Use one verified owner. The person represented controls the account, completes verification, and approves every published clip.
Set a narrow purpose. Begin with low-risk work such as tutorial introductions, event reminders, or alternate versions of a script you already approved.
Disclose the synthetic performance. Put a plain-language note in the video, caption, description, or nearby copy before a viewer has to guess.
Review the final export. Check the spoken claim, pronunciation, mouth movement, body motion, background, product details, and any on-screen text. Review the file, not just the prompt.
Keep a simple asset log. Record the script, generation date, platform, approver, disclosure used, published location, and deletion status. This makes corrections possible later.
Try it where mistakes are cheap
Your first test should not be a customer testimonial, a medical or financial explanation, an apology, or a message that depends on emotional authenticity. Make a short internal clip or an unpublished tutorial introduction. Compare the time spent writing, generating, correcting, and exporting with the time a normal recording would take.
Also test whether the result still feels like your brand. An avatar that saves 20 minutes but sounds stiff, mispronounces a product name, or makes your audience wonder what is real has not saved useful time. Keep the format only if the final clip is accurate, recognizable, and easy to disclose.
What to watch next
Check feature availability and limits in your own account. Workspace administrators can turn personal avatars off, and plan limits differ. Google also lets users retake or delete avatar data through their Google Account; the company notes that deletion affects other Google services that were allowed to use the same avatar, including the Gemini app.
A verified AI double can help a one-person business produce more video without arranging another shoot. Treat the avatar like access to your voice, not like another template. Give it an owner, a permitted use, a disclosure rule, and a final human review.
Sources
Google Docs Editors Help - Create, use and manage your personal avatar with Gemini in Google Vids
Google Docs Editors Help - Use AI to generate video clips in Google Vids
Google - Generative AI Prohibited Use Policy
Google - Making it easier to understand how content was created and edited
