Google’s Genie 4 Adds Real-Time Audio to Interactive AI Worlds

Representative image: AI-generated conceptual illustration.

Google DeepMind has introduced Genie 4, a research model designed to generate interactive worlds that users can steer with natural-language instructions. The company says the model creates sound alongside the scene, including ambient audio, music and dialogue.

In its October 27 announcement, Google said users can direct changes such as camera movement, weather and the introduction of characters while a simulation is running. The model draws on capabilities from the Gemini family to support visual realism and continuity, the company said.

From generated clips to environments

The distinction is interactivity. A conventional generated video plays through a fixed sequence; a world model aims to respond to new instructions as the scene unfolds. Google points to possible uses in creative production and as training environments for robots and software agents.

Those applications remain proposals rather than proof that the system is reliable enough for every production or safety-critical task. Simulated environments may still behave inconsistently or differ from the physical world, and Google's announcement does not establish independent performance across those uses.

Who can use it?

Google said the rollout had begun with a limited group of trusted testers. It plans access for Google AI Ultra subscribers through Project Genie and for early-access partners through the Gemini API and AI Studio. The announcement describes those wider channels as forthcoming, not universally available.

The research is a notable step toward AI-generated environments that respond to people in real time, but its practical value will depend on access, consistency and the results of independent testing.