Google rolled out a brand new experimental AI tool last Thursday called Project Genie. By Friday, video game stocks were tumbling as a result. Gaming industry giants like Unity Software, Roblox, Take-Two, and AppLovin all felt the effects of Project Genie, at least on Wall Street.
Project Genie, which is currently only available to subscribers to Google’s $249 per month AI Ultra plan, is a new generative AI world model from the company’s DeepMind research lab. Project Genie allows users to create interactive virtual worlds. Using nothing but text and image prompts, Project Genie users can create not just the environment, but also characters that interact realistically with the virtual space.
Looking at some examples of Project Genie in action, it’s easy to see why investors who are already bullish on AI would feel the same about this tool’s potential impacts on game developers.
At Reddit and X, users are trading examples of their favorite Project Genie virtual worlds:
This Tweet is currently unavailable. It might be loading or has been removed.
This Tweet is currently unavailable. It might be loading or has been removed.
This Tweet is currently unavailable. It might be loading or has been removed.
This Tweet is currently unavailable. It might be loading or has been removed.
This Tweet is currently unavailable. It might be loading or has been removed.
Riley Goodside, a staff prompt engineer at Google DeepMind, shared a video showcasing a 3D box of cigarettes coming to life as Goodside moves it around the floor of a subway station. According to Goodside, he provided the Genie 3 model with the prompts “34th Street-Penn Station” for the environment and “discarded pack of cigarettes” for the character, alongside an initial frame of the scene generated with Nano Banana Pro.
Mashable Light Speed
This Tweet is currently unavailable. It might be loading or has been removed.
Goodside explained that users have plenty of control over what’s generated via the image used for the initial frame, but less control over the environment once the user moves their character around the virtual world.
Some of those limitations become more obvious in other examples, like this Project Genie-generated creation of a character attempting to look in the mirror.
This Tweet is currently unavailable. It might be loading or has been removed.
Project Genie also seems to treat secondary characters as just inanimate objects in some other examples shared to the social media platform X.
This Tweet is currently unavailable. It might be loading or has been removed.
This Tweet is currently unavailable. It might be loading or has been removed.
And much like other generative AI tools, it seems like there will be clear-cut copyright issues with some of the content being generated.
This Tweet is currently unavailable. It might be loading or has been removed.
Some industry heads, like Unity CEO Matthew Bromberg, don’t seem too concerned about world models replacing game engines, with Bromberg making the case that they will enhance output from experienced game developers.
This Tweet is currently unavailable. It might be loading or has been removed.
Still, it’s very early days for Project Genie, and it’s already able to generate some detailed “worlds.”
Disclosure: Ziff Davis, Mashable’s parent company, in April 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.
Topics
Artificial Intelligence
