Google has debuted Project Genie, an experimental research prototype that lets Google AI Ultra subscribers in the United States create, explore, and remix interactive virtual environments using simple text and image prompts. Powered by DeepMind’s Genie 3 world model, along with Google’s Nano Banana Pro image generator and Gemini AI, the tool generates dynamic 3D worlds that respond to user movement and actions in real time. Through features like World Sketching, users can design landscapes and characters, preview environments, then navigate them in first- or third-person perspectives. The prototype also lets users remix existing worlds and download short videos of their experiences. Google underscores that Project Genie is still early research: environments may not perfectly match prompts, character controls and realism are limited, and sessions currently cap at around 60 seconds. Broader access beyond U.S. Ultra subscribers is planned as the technology develops.

Source:  
https://blog.google/innovation-and-ai/models-and-research/google-deepmind/project-genie/