Google grounds Project Genie worlds in Street View imagery
Google's Project Genie now anchors generated worlds in Street View imagery, rolling out to $200 Google AI Ultra subscribers and grounded in tech Waymo already uses.

Updated
Why it matters
- Project Genie now supports Street View grounding via Maps Imagery Grounding for U.S. locations, with expansion planned.
- The feature is gradually rolling out to eligible Google AI Ultra subscribers globally, 18+, on the $200 plan.
- Google says Genie already serves as a research tool and helps Waymo simulate hyper-realistic road environments.
Google has connected Project Genie, its experimental world-generation prototype, to Street View, letting users build interactive environments anchored to real-world imagery of places in the United States. The capability, called Maps Imagery Grounding, rolls out gradually starting today to eligible Google AI Ultra subscribers globally, aged 18 and over, on the $200-per-month plan.
The stakes go well than novelty world-building. Google describes Genie as a general-purpose world model that has already become what the company calls "a foundational tool for research, enabling agents to learn and reason in complex virtual settings and even helping Waymo simulate hyper-realistic road environments." Grounding generated worlds in Street View imagery means AI agents and robots could train and reason inside simulations that correspond to actual places — a step toward closing the gap between virtual training environments and the physical world.
How the feature works
Inside Project Genie, users tap a Maps pin to select a place in the U.S. They can optionally apply a visual style — Google lists "Desert Sands," "Stone Age," "B&W film" and "Ocean World" as examples — and then describe a character, such as a favorite animal, a comic book hero or a claymation monster. Genie uses those inputs to generate an imaginative world whose starting location is tied to Street View's real-world imagery.
Google offers two concrete examples. Selecting "Ocean World" lets a user scuba dive with schools of fish around the Golden Gate Bridge. Selecting "B&W film" at the Fort Worth Stockyards in Texas produces a version of the location as it might have looked in the 1920s, complete with saloons, vintage cars and trading posts.
The underlying technology is Maps Imagery Grounding — the same system Google says developers already use to create AI visuals with Street View. Google has not announced expansion beyond the U.S., but the company says Street View imagery in Project Genie "is available now for places in the U.S. with plans to expand to more places over time."
Availability and status
Project Genie, including the Street View capability, is gradually rolling out to all eligible Google AI Ultra subscribers globally, 18 and older. Google is also broadening access to Project Genie to more people around the world as part of this release.
Google frames Project Genie as an experimental research prototype inside Google Labs, and it flags unfinished edges. The team says it is "working behind the scenes to make the details even sharper and more accurate," and points users to its website for current limitations.
Why it matters
The release ties two threads of Google's AI strategy together. The first is world models — systems that generate interactive, navigable environments rather than static images or text. Genie already serves that role in research settings, and its application at Waymo shows how generated environments can support training for autonomous systems that must operate in the real world.
The second thread is grounding: connecting generative models to real geographic data so their outputs correspond to places that actually exist. By anchoring generated worlds in Street View imagery, Google gives agents — and eventually robots, as the company suggests — a simulation substrate tied to real locations rather than invented ones.
For now, the consumer-facing product is playful: dive the Golden Gate Bridge, tour a 1920s Texas stockyards, drop a claymation monster into a stylized desert. But the research direction is explicit in Google's own framing, which says the expansion "can provide a virtual environment for AI agents or robots to navigate and interact with the complexities of the real world." The price of entry — a $200 Google AI Ultra subscription — also signals where Google places this work within its premium AI portfolio.
The immediate next markers to watch are geographic expansion beyond the U.S., improvements in visual fidelity that Google says are underway, and whether Maps Imagery Grounding moves from a consumer experiment into the agent-training pipeline that Genie already feeds at Waymo.
Original: waymo.com
More from Sophie Lindqvist
Show full bio
Staff writer covering marketplaces and e-commerce at AI In Context.
114 articles
Related articles
- Google Opens Project Genie to AI Ultra Subscribers in the U.S.
- Google DeepMind's SIMA 2 Turns AI Into a Gaming Companion
- Google DeepMind Launches AI Accelerator for Climate in Asia-Pacific
- Google DeepMind Launches European Robotics Accelerator
- Google DeepMind Opens Singapore Research Lab to Serve Asia-Pacific AI Push