← All AI news
Google DeepMindIndustry

Genie 4 brings sound and live direction to simulated worlds

Google’s new world-model research combines interactive scenes with native audio, with access beginning through a limited tester rollout.

THE UPDATE

What changed

Google announced Genie 4, a world model for interactive simulation. People can steer a generated scene with natural-language directions, while the model produces audio alongside the visuals. Google says a limited group of trusted testers has begun receiving access; broader routes are planned through Project Genie and early-access developer partners.

One useful next step

Write a short simulation brief: the situation, what a person should be able to change, and three behaviors you would need to verify. Use that brief to assess demos before planning a production dependency.

Start with a clear project brief ↗

Before you dive in

The announcement describes a limited rollout. Planned availability is not the same as access in your account today, and a research demo is not evidence of reliability for a specific real-world task.

GO TO THE ORIGINAL

Source & announcement date

Google DeepMind ·

Google: introducing Genie 4 ↗ (opens in a new tab)

This brief summarizes the linked announcement. “The AI Superpower take” and the next step are our editorial interpretation, not claims from the source.

KEEP YOUR CURIOSITY GOING

A few more worth knowing.

All AI news ↗