Google DeepMind has introduced Project Genie, a new research tool that lets users create and explore virtual worlds in real time. The project is based on the Genie 3 world model and is currently available to Google AI Ultra subscribers in the United States who are 18 years or older.
Project Genie is different from normal AI tools that only make images or videos. Instead, it creates worlds that users can move through and interact with. As a person walks, flies, or drives inside the world, the AI builds the environment ahead of them instantly.
The tool works in three main ways. First, users can create worlds using text prompts or images they upload or generate. These worlds can be edited and adjusted before entering them. Second, users can explore in real time. There is no fixed map—what lies ahead is created as the user moves. The experience can be viewed from a first-person or third-person angle. Third, users can remix worlds by taking existing environments and changing prompts to create new versions.
Because Project Genie is still in an early testing stage, it has several limits. Each session only lasts 60 seconds due to the high computing power needed. The visuals run at 720p resolution with a frame rate of about 20 to 24 frames per second. Google says movement may sometimes feel delayed, physics may not work perfectly, and the worlds may not always look realistic. Some planned features, such as events that change the world during exploration, are not included yet.
Public reaction on X (Twitter) to Project Genie has been honest and full of feedback and surprise.
One user who tested the tool said it did not run as smoothly as the marketing videos, mentioning lag, short time limits, and trouble using custom prompts. However, the same user still praised the technology, saying the worlds felt more stable and realistic than earlier tools, and that the experience showed real progress despite the flaws.
https://x.com/yrzhe_top/status/2016943806298382523
Other users reacted with a joke that AI tools like this might make it possible to build a game like GTA 6 before it is officially released.
https://x.com/masterzorgon/status/2017003461141770527
Another pointed out how fast AI has grown, moving from simple text generation to full, explorable worlds in just a few years, and wondered how game developers might use this technology very soon.
https://x.com/praveenyen/status/2017105523322343563
Google DeepMind says Project Genie is part of its long-term work on building AI systems that understand how the real world works. These systems could help train AI used in areas like robotics, where machines need to learn cause and effect. In the future, similar technology could also be used for making games, films, or digital history experiences.
For now, Project Genie is a research experiment. But it shows how Google DeepMind is moving toward AI that can create and respond to entire virtual environments, not just single images or scenes.
