Google Gemini can present cartoon or lifelike avatars to lip-sync generative chatter

Google has introduced a new feature called Live Avatar to its Gemini 3.8 Live model, enabling the creation of animated characters and realistic human simulacra that can lip-sync machine-generated speech. This multimodal capability allows for more engaging conversations by processing both visual and audio inputs simultaneously. The feature, explained by Google DeepMind researchers Shuo-yiin Chang and CJ Zheng, is designed to enhance enterprise agents' conversational experience, catering to various applications such as hotel guest check-ins.

The introduction of Live Avatar comes at a time when concerns about anthropomorphizing AI models are surfacing, with potential risks including inflated user expectations and reliance on conversational agents. Despite these concerns, Google has implemented strict safeguards to ensure transparency and protect user identity while deploying these AI-generated personas across multiple languages.