On paper, Project Genie is a dream toy for those curious about new technologies. Developed by DeepMind, the tool allows you to create interactive worlds from a simple text description, possibly accompanied by images. All you have to do is describe a setting, a character, an atmosphere, and the AI takes care of creating a world in which you can move around.
AI invites itself into known licenses
The particularity of Genie 3, the model that powers Project Genie, is that it is not satisfied with a fixed scene. The world is built through exploration. With each step, with each camera movement, the environment expands, adapts and reacts, with a form of simulated physics. You can walk, drive, fly, switch from third-person perspective, and even remix existing worlds to create new ones.
Subscribe to WorldOfSoftware
Google, however, prefers to calm the enthusiasm. The company emphasizes the experimental nature of the project and its current limits. The worlds generated are not always very credible, the instructions are not systematically followed to the letter, and the laws of physics can be interpreted with a certain freedom. The characters sometimes lack precision in the controls, and latency can be felt.
Another constraint: sessions are limited to 60 seconds. Currently, Project Genie is limited to Google AI Ultra subscribers in the United States and only to users over 18 years old. In short, we are still far from a general public tool ready to replace a traditional game engine.
Where things get complicated is when Project Genie leaves the field of simple technical experimentation to enter that of cultural creation. In a presentation video, Google explains that the user can “be any character”. A sentence that quickly raised eyebrows.
We can already find generated worlds on social networks featuring well-known video game characters and worlds. Creations that immediately recall protected licenses (Mario, Zelda, Metroid…), and which raise an obvious question: how far can we go without encroaching on copyright?
Twitter-tweet” data-width=”500″ data-dnt=”true”>
Holy moly… Genie 3 just created this mock 3D game world from Breath of the Wild.
How I did it + prompts in comment. https://t.co/IrwlZ1pTMs pic.twitter.com/H33an42YNd
— Min Choi (@minchoi) January 29, 2026
Project Genie only concentrates problems already well identified with generative AI. What data was the model trained on? Was this data usable in this context? And above all, what happens when these tools make it possible to recreate universes likely to compete with existing works? Generating an image or meme inspired by a famous character is one thing. Creating an entire playable world, potentially reusable or redistributable, is another. Google, as of yet, does not mention any such explicit restrictions for Project Genie.
🟣 To not miss any news on the WorldOfSoftware, follow us on Google and on our WhatsApp channel. And if you love us, .
