Google's Genie world model now simulates real streets with Street View

At a glance:

• Google has integrated Street View with Project Genie, allowing users to simulate real-world streets with adjustable conditions
• The feature launched at Google I/O 2026 and is being used by Waymo to train self-driving cars on rare events
• Genie 3 can create interactive environments from text prompts but remains experimental with limitations in physics awareness

What is Google's Genie-Street View integration?

Google DeepMind has connected Street View to Project Genie, creating a powerful new capability that allows users to simulate real-world streets in an immersive, interactive way. This integration enables people to explore familiar locations while adjusting conditions like weather or time of day, essentially creating virtual environments that mirror real places. The new feature was officially announced during the Google I/O 2026 developer conference, marking a significant advancement in how AI can leverage Google's extensive mapping data.

"This is really powerful for both the agent [and robotics] use case and for humans to play with, and that's always been the thesis of Genie," explained Jack Parker-Holder, a research scientist on DeepMind's open-endedness team. He provided concrete examples of how this technology could be applied, such as simulating rare sunny moments in London for robots that typically operate in overcast conditions, or allowing users to see what a New York City neighborhood would look like during a snowstorm.

How does Street View integration work?

Google has been collecting Street View data for two decades using specialized camera-equipped vehicles and individuals wearing "tracker backpacks." This massive dataset includes over 280 billion images captured across 110 countries and all seven continents, providing an unprecedented wealth of real-world visual information. By combining this rich repository of actual street imagery with Genie's generative capabilities, Google can create simulations that are grounded in real locations while allowing for creative manipulation of environmental factors.

The integration works by taking the extensive Street View imagery and using Genie 3 to generate interactive environments based on these real-world references. Users can input prompts or select locations to see how they might look under different conditions. This approach differs from traditional simulation tools that might create entirely fictional environments, instead offering a unique blend of authentic location data with AI-powered environmental modifications.

Who benefits from this technology?

The Genie-Street View integration serves multiple use cases across different industries and user groups. For robotics, it provides valuable training environments where robots can experience conditions they might rarely encounter in the real world, such as unusual lighting or weather patterns. Waymo, Google's autonomous vehicle subsidiary, is already leveraging this technology to train its self-driving cars on "exceedingly rare events" like tornadoes or unexpected animal encounters.

For consumers and developers, the technology opens up new possibilities for exploration and creativity. "With Street View, we have imagery from a large quantity of the world," Parker-Holder noted. "You can imagine how potentially powerful it is to combine this rich source of real-world information and data with an ability to simulate worlds." This could lead to new forms of education, entertainment, and virtual tourism that blend authenticity with imaginative possibilities.

How is this different from previous simulation tools?

Unlike traditional simulation tools that might be limited to a single perspective or entirely fictional environments, Genie's integration with Street View offers several unique advantages. According to Parker-Holder, Waymo's existing simulator focuses primarily from the car's point of view, while Genie allows for shifting perspectives to different types of agents, including humans or specialized robots.

The key breakthrough lies in the AI's spatial continuity. As Jonathan Herbert, director of Google Maps who started on the Street View team as an intern 12 years ago, explained: "Genie can't yet create a faithful reconstruction of a street. I think the real breakthrough is the AI's spatial continuity. If you turn 360 degrees, the AI correctly remembers and simulates the environment behind you. From that point on, the model can build a new environment on top of that." This ability to maintain context while exploring environments represents a significant advancement in spatial AI.

What are the current limitations?

Despite its impressive capabilities, the Genie-Street View integration remains an experimental technology with notable limitations. The models currently produce results that are "impressive and recognizable, but still video game quality rather than photorealistic," according to Parker-Holder. More significantly, the system is not yet physics-aware, meaning it doesn't understand cause and effect relationships in simulated environments.

In demonstrations, the AI has shown limitations such as allowing characters to run through solid objects like cacti and bushes without physical interaction. This contrasts with more advanced AI systems like Google's image generator Nano Banana, which can generate perfect text in infographics, or its video generator Veo, which understands physical phenomena like paper boats drifting on water currents or smoke dispersing into air. Parker-Holder acknowledged this gap, noting: "I think for this kind of model, it's maybe six to 12 months behind video in terms of the accuracy and quality, so I think it's something we will solve."

What's next for Genie and Street View?

Google is currently rolling out the Street View integration in Genie to select Ultra users in the United States, with plans to expand access globally over the coming weeks. The company's researchers are focused on improving the technology's accuracy and physical realism as they continue development.

Diego Rivas, a product manager at DeepMind, emphasized that while the technology shows promise, it remains experimental: "We want to put this new capability into as many hands as possible," he said, "but Street View in particular and Genie in general is still an experiment, so there's much to improve upon in terms of accuracy." The long-term vision extends beyond entertainment and training, potentially revolutionizing how humans interact with and understand geographic spaces through AI-powered simulation.