Google has turned its Gemini personal avatar into a shared creation tool, letting friends and collaborators place a person’s face and voice into AI-generated images and videos once that person agrees. The avatar is built from a short recording of the owner’s face and voice, and the shared version works through a tag in a prompt, rather than requiring everyone to record together. Images are generated with Nano Banana, videos with Gemini Omni, and video creation requires a Google AI plan.
What changed
The change is the sharing step, not the avatar itself. Google let an eligible user record a face and voice sample to create a personal avatar for images and videos. Sharing extends that avatar to other people, so a friend who accepts an invitation can tag the owner in a prompt and generate content featuring them, together in one scene.
Google announced the sharing step on 29 October through its Gemini app account, and its support page now describes the process. The move follows a busy week for the assistant, which also gained in-chat file editing that week. Together, the two updates shift Gemini from a solo chat tool toward a shared workspace where other people can enter the result.
How sharing works
Creating an avatar starts on a computer, where Gemini shows a QR code that the owner scans with a phone or tablet. The owner then records a short video of their face while speaking, following on-screen guidance on lighting, framing and background. Google says the account owner must complete the recording, and minors must not be recorded for an avatar.
Sharing is done by invitation to a Gmail address, and the recipient must accept before using the avatar. Once accepted, the recipient can tag the owner in a prompt, for example by asking for an image of themselves dancing with the owner. The owner can set an expiry date for access, remove a person, or retake the recording.
What stays limited
Eligibility is narrower than a general Gemini feature. Users must be at least 18, signed in with a personal Google account, and using English. Avatars are not available in the United Kingdom, the European Economic Area or Switzerland, and creating a video with an avatar requires a Google AI plan, while images do not carry that extra requirement.
The controls also have a limit that owners should understand. Removing a person’s access stops them from creating new images or videos with the avatar, but it does not delete anything they made. Deleting the avatar removes the recorded selfie and voice data from Google’s systems, yet videos and images already made and published with it are not deleted.
Our read: consent is the product here
Our read is that consent, not image quality, is the point of this release. A shared likeness works if the person agreed to be there, and Google has built that agreement into the workflow through invitations, acceptance, expiry dates and revocation. That design matters more than any single scene, because trust lets friends use each other’s faces at all.
The weakness is the one Google states: revocation is forward-looking only. An owner who changes their mind can stop future creations, but cannot pull back images or videos shared or published. Anyone considering an avatar should treat sharing as a lasting decision, even though access itself can be withdrawn, and should share first with people whose judgement they trust.
Why It Matters
For users, the practical gain is group creativity without recording sessions. Friends, family members and collaborators can appear together in images and videos remotely, with character and voice consistency carried by the avatar rather than by repeated filming. That convenience is real, and it is also why the permission controls deserve more attention than the novelty of the first picture.
For the wider market, shared avatars raise the same likeness questions that new deepfake safeguards are trying to address, even though Google’s version is consent-based. A standard cannot settle every dispute over a shared face or voice, but it shows where the pressure is heading: toward systems that can demonstrate permission, limits and removal, not merely generate a convincing result.
Background
Personal avatars in Gemini arrived earlier in 2026, before the sharing step opened them to collaborative use on 29 October. Google’s support page frames the feature as a way to streamline creation, while its privacy hub covers how avatar recordings are handled. The sharing announcement arrived alongside other Gemini updates that week, as Google expanded the assistant beyond question-and-answer chat.
Source: Google support page
0 Comments