Google rolls out Gemini avatar for AI video clones

08:54

By: Dakir Madiha

Google has expanded access to its Gemini Avatar feature, a tool that allows paid subscribers to generate realistic AI video versions of themselves using their own face and voice. The system runs on the company’s Gemini Omni model and is being released more broadly after its first presentation at Google I/O 2026 on May 19.

The setup process takes only a few minutes. Users open the avatar settings inside the Gemini application and complete a guided onboarding sequence using the phone’s front camera. The system asks users to face the camera, turn their head, and read random numbers aloud. This step captures facial structure and voice patterns to build a digital replica linked to the account.

Once created, the avatar can be used inside Gemini conversations by typing a personal tag or the command @me. The generated outputs aim for high realism in facial movement and voice reproduction. Early testers report that the results can closely resemble real speech patterns and expressions, enough to appear convincing even in casual viewing contexts.

The feature raises immediate concerns about synthetic media and misuse. Google requires users to be at least 18 years old and enforces live presence during setup to prevent the creation of avatars using stolen images or recordings. Each generated video includes a hidden SynthID watermark designed by Google DeepMind, which can be verified across Google services to track AI-generated content.

Additional safeguards limit broader manipulation tools. Voice editing and advanced audio transformation features remain restricted while the company evaluates safe deployment. The system arrives at a time when deepfake risks are increasing across digital platforms, with regulators and technology companies under pressure to control realistic synthetic content.

Availability is currently limited to paid Google AI Plus, Pro, and Ultra subscribers in the United States and select regions. Access is not yet extended to parts of Europe, including the United Kingdom and Switzerland, reflecting ongoing regulatory and policy differences around generative AI tools.