Gemini will soon let you build an AI avatar of your own face and voice

AI AVATARSGEMINIMy Avatar in GeminiTECHPLUGGED.COM

Google is putting a face-and-voice cloner inside the Gemini app. In a notice to Workspace administrators, the company said a new My Avatar feature will let users build an avatar from their own face and voice, arriving no sooner than August 4, 2026.

The pitch is that it happens “safely and securely,” and Google has drawn some fairly firm lines around it. Only users can create, manage and delete their own avatar — an admin can kill the feature entirely but cannot reach in and make one on someone’s behalf, or delete theirs. That distinction matters more than it sounds. A corporate avatar library that IT controls would be a governance nightmare and a deepfake vector wrapped in the same box; keeping the likeness in the hands of its owner is the right call.

Adults only, English only

Eligibility is tight at launch. Users must be at least 18 and have English set as their primary language. Both the Gemini app and the My Avatar setting have to be enabled by the organisation for the feature to appear at all — if an org has Gemini switched off, the avatar toggle will still read “ON” in the Admin console but nothing will surface to users.

The age gate is the obvious tell. Synthetic likenesses of minors are a line no major platform wants to be caught on the wrong side of, and restricting to over-18s is the cheapest way to stay clear of it. The English-only limit reads more like a voice-model capability constraint than a policy one, and will presumably lift as coverage improves.

On by default — again

As of July 16, 2026, the avatar setting is ON by default for all eligible customers, with the control rolling out to every Admin console by July 21. Organisations that would rather not have staff minting synthetic versions of themselves need to go to Generative AI > Gemini app > My Avatar in Gemini, uncheck it, and save. Google notes admins can verify the setting’s status via the Gemini app audit log.

It’s the second opt-out-by-default AI rollout Google has told Workspace admins about this week, alongside Google Pics becoming a core service in August. The pattern is unmistakable: ship it enabled, let inertia do the adoption work, and put the burden of saying no on the customer.

The unanswered question is what avatars are actually for here. Google’s notice describes the mechanics in detail and the use case not at all — no mention of Meet, no mention of recorded updates, no mention of whether your avatar can show up anywhere other than the Gemini app. A cloned face and voice with no stated destination is a feature looking for a job. Expect the answer to arrive on the Workspace Updates Blog rather than in an admin email.