Google connects its Genie world model to Street View imagery, letting users generate interactive AI environments anchored to real locations

Google Deepmind is rolling out Genie 3 with Street View grounding globally to Google AI Ultra subscribers ($200 per month, 18+). Users drop a pin on a map, select an optional style like 'Ocean World' or 'Desert Sands,' describe a character, and Genie builds a walkable world whose starting point is tied to real Street View footage. The real-world location feature currently works only for U.S. spots, with plans to expand.

The feature relies on 'Maps Imagery Grounding,' an interface developers already use to create AI-powered visuals from Street View data. Google's Street View library serves as both training material and an anchor for these generative worlds, giving the company an advantage in grounding AI environments to actual places.

The world model is designed primarily to give AI agents and robots a training environment where they can navigate, reason, and learn. Deepmind's agent SIMA 2 already uses Genie as a training ground, and Waymo uses it to simulate realistic street scenarios for self-driving cars. The Street View connection means these training runs can now be pinned to specific real-world locations.

Deepmind describes this as an experimental research prototype and says the team is working on a better version. Early demos show soft textures, unstable geometry, and surreal transitions still present in the generated environments.