Google has added a "Live Avatar" feature to Gemini 3.8 Live, enabling animated or photorealistic avatars to lip-sync the AI's machine-generated speech. This development raises significant ethical concerns because Google's own research—spanning from a 2021 DeepMind paper to a December 2025 study—has warned that making AI appear human-like can mislead users into overestimating the system's capabilities and placing unwarranted trust in it.

The Live Avatar feature can use pre-made character designs or custom avatars, supports conversation in 97 languages, and is suited for enterprise scenarios such as hotel guest check-ins. The tension between Google's deployment decision and its own research is noteworthy: in August 2025, OpenAI faced a lawsuit from parents alleging that ChatGPT's "anthropomorphic mannerisms" contributed to their son's suicide, yet Google is proceeding with this feature despite its researchers' documented concerns that humanlike presentation fosters false confidence and can obscure AI errors.