Google Gemini users can now record their face and voice to create personal avatars for AI-generated content. The idea is simple to describe and striking in its implications: instead of a generic AI character, the content can feature a likeness built from the user themselves. Sharing is handled through Gmail invitations, with recipients needing approval before they can use an avatar, and owners keep control through Gemini settings.
The starting point is a recording. Gemini users can record their face and voice to create a personal avatar, which can then appear in AI-generated content. The avatar is personal in a literal sense: it is built from the individual user, not from a stock character or a celebrity likeness, and it exists to represent that user inside generated media.
That personal quality is what makes the feature different from generic AI video tools. A generic tool asks the user to pick a character. This feature asks the user to become the character, in a controlled way, for content they choose to make. The distinction matters for trust, because viewers will reasonably ask whether the person on screen is really the person, and in what context the avatar was used.
Sharing works through Gmail invitations. Users can share their avatars with other Google users through Gmail invitations, and recipients need approval before they can use them. That approval step is important: sharing is not the same as giving away control, and the recipient cannot simply take an avatar and use it without the owner saying yes.
Avatar owners can manage permissions, set expiry dates, and revoke access through Gemini settings. Each of those controls answers a different worry. Permissions decide who can do what. Expiry dates put a time limit on access, so a collaboration does not last forever by default. Revocation gives the owner a way to end access when a project ends or a relationship changes.
Access comes with clear limits. The feature requires a personal Google Account, and it is limited to users aged 18 and above. Those two conditions together draw a line around who the feature is for: individual adults using their own accounts, not younger users and not, on the reported terms, other account types.
Language and geography are limited as well. The feature supports English, and it excludes the UK, EEA, and Switzerland. Those exclusions are significant, because they remove several large markets from the initial picture. Anyone planning content around personal avatars needs to check whether their audience is inside the supported scope before assuming the feature will reach them.
Finally, video generation requires a Google AI plan. That condition separates the basic idea of creating an avatar from the more demanding job of generating video with it. It is a reminder that AI video remains a premium capability, with costs and infrastructure behind it that a free tier does not always cover.
Personal avatars sit at the intersection of content and identity, and that is why SEOs should pay attention even though the feature is not a ranking factor. Search and AI systems are increasingly asked to judge whether content is authentic, and a world of personal avatars makes that judgment harder and more important at the same time.
The approval and expiry controls are the most transferable lesson. As AI-generated content becomes easier to make, the tools that give creators control over their likeness will be the ones brands can safely use. A feature without revocation is a liability. A feature with it is a workflow.
For link builders, the near-term effect is indirect. Avatar content may earn attention, and attention can earn links, but only if the content itself is worth citing. An avatar does not make a thin idea interesting. It makes an interesting idea easier to present, and that distinction should guide how the feature is used.
Watch how audiences react to avatar content once it circulates widely. Early reactions will set norms: when viewers accept an avatar as a legitimate stand-in, and when they expect a real person instead. Those norms will shape brand guidelines faster than any platform policy.
Watch, too, whether the geographic and language limits change over time. An English-only feature that excludes the UK, EEA, and Switzerland is a limited launch by design, and its expansion, or lack of it, will signal how Google weighs the risks and rewards of personal likeness tools.
Source: Gadgets 360, October 30, 2026.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial