Key Highlights:

• The project signals Google’s next step toward advanced world models and AGI research.
• Google has opened access to Project Genie, an experimental interactive world-creation tool.
• The prototype is powered by Genie 3 and is available to Google AI Ultra subscribers in the US.
• Users can create, explore, and remix infinite interactive worlds in real time.

Google has begun rolling out Project Genie, a research prototype that lets users create and explore infinite, interactive worlds using AI. Built on the Genie 3 world model, the project matters because it shows how AI systems could soon simulate complex environments in real time, a key step toward more general-purpose artificial intelligence.

Project Genie is now available to Google AI Ultra subscribers in the US aged 18 and above. The tool lives inside Google Labs and offers early public access to technology that was previously limited to trusted testers.

What exactly is Project Genie?

Project Genie is an experimental web-based prototype developed by Google DeepMind. It is powered by Genie 3, along with Nano Banana Pro and Gemini.

At its core, Project Genie allows users to generate living, interactive worlds that respond to movement and actions in real time. Unlike static 3D scenes, the environment unfolds dynamically as users explore it.

How does the model work?

A world model predicts how an environment changes over time and how actions affect it. According to Google, this capability is critical for building systems that can operate in the real world, not just in narrow tasks like chess or Go.

Genie 3 generates the path ahead as users move through a scene. It simulates physics and interactions while maintaining consistency across the world. This makes it useful for areas such as robotics research, animation, fictional storytelling, and recreating real or historical locations.

What can users do?

Project Genie focuses on three main capabilities. First is world sketching. Users can prompt the system with text, images, or uploaded visuals to create an expanding environment. They can define characters, movement styles, and camera perspectives before entering the world.

Second is world exploration. The environment is fully navigable. As users move, the system generates new terrain and interactions in real time.

Third is world remixing. Users can build on existing worlds, explore curated examples, or remix prompts to create new interpretations. Completed sessions can be exported as videos.

What are the current limitations?

Google says Project Genie is still an early research prototype. Generated worlds may not always match prompts or real-world physics. Character control can feel inconsistent, and sessions are currently limited to 60 seconds. Some features announced earlier, such as prompt-driven world events, are not yet included.

Why this matters going forward?

Project Genie gives a public glimpse into how world models could shape future AI research and generative media. As Google continues testing Genie 3 through this prototype, the Project will likely play a key role in understanding how people interact with simulated worlds at scale.