Google Rolls Out Project Genie, Letting Users Generate Playable AI Worlds

Google DeepMind launched an experimental prototype called Project Genie on January 29, 2026, which enables users to create and explore interactive virtual worlds powered by artificial intelligence.

Credit: Google

Project Genie, available to subscribers of Google AI Ultra in the United States who are at least 18 years old, generates 3D environments from text prompts or images and allows navigation through them in real time.

It builds on Genie 3, a world model that simulates physics and continuity as users move characters via keyboard controls, such as WASD keys for movement and the space bar for jumping.

Sessions last up to 60 seconds at 720p resolution and 24 frames per second, with worlds expanding dynamically during exploration.

Users can remix existing worlds or start new ones, but the system blocks certain prompts involving third-party content.

"In August, we previewed Genie 3, a general-purpose world model capable of generating diverse, interactive environments," Google DeepMind stated in its announcement.

The tool is capable of generating virtually infinite playable worlds, just like we play video games and users can generate these worlds from personal photos too.

Diego Rivas, a product manager at Google DeepMind, said of the prototype, “It’s really for us to actually learn about new use cases that we hadn’t thought about.”

The release follows a limited research preview from the previous year and aims to gather feedback on applications like visualizing scenes for films or creating educational tools.

Google plans to expand availability beyond the United States in the coming months.

Early tests showed the tool producing environments inspired by games, though with limitations like input lag and no audio.