Google DeepMind Connects Street View to Project Genie World Model
Google DeepMind has connected its Project Genie world model to Google Street View’s 280 billion images, letting users explore AI-generated simulations of real locations. Announced at Google I/O, the feature is rolling out to Google AI Ultra subscribers in the US, with Waymo already using Genie 3 to train self-driving cars on rare scenarios.
Integrating Generative World Models and Real-World Data
DeepMind’s Project Genie, its general-purpose system for creating interactive environments, now leverages more than 280 billion images captured across 110 countries and all seven continents. This integration allows users to navigate AI-generated simulations of real places in real time, such as a snowy New York City block or a sunlit London street.
From Research Preview to Consumer Product
Genie 3, the latest iteration, initially appeared as a research preview in August 2025. In January 2026, DeepMind opened access to Google AI Ultra subscribers in the US. The Street View integration is now available to some Ultra users in the US, with global expansion planned for the coming weeks.
Applications and Impact
The feature serves two primary audiences: robotics developers who use it to train agents in simulated environments mirroring actual locations, and general users who can explore for fun. Waymo already utilizes Genie 3 in its simulators to train on rare events, enhancing the realism of their simulations by grounding them in real-world Street View data.
This simulation-to-reality pipeline is crucial for bridging the gap between what robots learn inside computers and how they perform once deployed. Companies like Nvidia and Cadence are also racing to achieve this, with DeepMind’s approach offering a distinct path by layering generative models on top of real-world imagery.
Limitations
While impressive, the Street View integration remains experimental. The generated environments resemble video games more than photographs, and the model lacks physics awareness.