Google DeepMind unveils 'Project Genie' powered by Genie 3 model

Project Genie, a new experimental research prototype from Google DeepMind unveiled on Thursday, uses its latest world-model artificial intelligence (AI) to develop and explore interactive virtual environments, according to a blog post on Google’s Innovation and AI site.

Powered by Genie 3, Project Genie is a general-purpose world model that can create diverse, explorable worlds from simple text and image prompts. Users can generate dynamic landscapes, characters, and environments that evolve in real time. These interactive elements are designed to respond directly to movement and actions.

The new project is a part of Google’s wider research into advanced AI systems that go beyond static text or image generation toward dynamic “world” building. Users can generate a wide array of simulations based on their descriptions, from natural environments such as deserts and forests to intricate ecosystems and imaginative, fantastical settings.

Google is now granting access to Project Genie for Google AI Ultra subscribers in the United States, enabling them to explore its world-generation capabilities.

The core technology is Genie 3, a breakthrough “world model” introduced in 2025. It represents a significant advance over earlier, short-lived scene generators by creating interactive environments that maintain logic and continuity through several minutes of user exploration.

Amid intense competition in generative AI, Project Genie arrives as AI companies such as OpenAI and Meta advance systems that support dynamic content creation.

Researchers consider the development of world models, such as Genie, as an important advancement toward more general AI capable of learning and reasoning in simulated settings.

Simultaneously, the AI sector is increasingly concentrating on spatial intelligence—a technical domain emphasizing an AI’s capacity to comprehend and create three-dimensional environments.