Waymo has announced the 'Waymo World Model,' an AI model that generates simulations for self-driving cars, including situations where elephants and dinosaurs appear on the road, based on Google's AI tool 'Genie 3.'

Waymo, a self-driving car development company under Google, has announced the ' Waymo World Model ,' an AI model that can create a 'super-realistic' simulation environment to train self-driving models.

The Waymo World Model: A New Frontier For Autonomous Driving Simulation
https://waymo.com/blog/2026/02/the-waymo-world-model-a-new-frontier-for-autonomous-driving-simulation

Waymo leverages Genie 3 to create a world model for self-driving cars - Ars Technica
https://arstechnica.com/google/2026/02/waymo-leverages-genie-3-to-create-a-world-model-for-self-driving-cars/

What happens when Waymo runs into a tornado? Or an elephant? | The Verge
https://www.theverge.com/ad/857759/how-the-age-of-ai-is-making-human-expertise-more-valuable

The Waymo World Model is built on Genie 3 , a Google DeepMind AI model that creates 3D environments that users can navigate simply by entering text. When Google expanded access to this technology as Project Genie, it attracted enough attention to cause stock prices of gaming companies to fall.

Immediately after Google announced the global model 'Project Genie,' stock prices of game companies fell across the board - GIGAZINE

Waymo will build the Waymo World Model on top of Genie 3 to train its autonomous driving models in unusual situations that cannot be imagined in the real world.

'Most simulation models in the autonomous driving industry are trained from scratch based solely on collected road data. This approach limits the system to learning from limited experience. Genie 3 has strong world knowledge gained from pre-training on a very large and diverse set of videos. This enables the vehicle to explore situations it has never directly observed,' Waymo said.

For example, if a tornado appears ahead, the Waymo World Model will convert the vast amount of information obtained from 2D video into 3D information through specialized post-event learning, and will be trained to understand things like depth.


A scene in which a vehicle in front of a driver goes over the side of the road due to reckless driving.


A scene where an elephant stands in front of you.


It can also be trained to simulate rare situations that rarely occur in the real world, such as encountering a pedestrian disguised as a dinosaur.


By creating a 3D environment, it will be possible to simulate for a longer period of time than before, ensuring stricter safety standards.