Google has introduced an AI model called Gemini Nano in Chrome, which automatically installs on Windows, Mac, and Linux devices without explicitly notifying the user. This model, about 4 GB in size, went unnoticed by most until several users detected a file named weights.bin in the browser's folder, raising concerns and questions about its origin. Gemini Nano downloads in the background if the computer meets certain requirements: at least 16 GB of RAM, a four-core processor, a GPU with 6 GB of vRAM, and 22 GB of free space. It does not require user intervention, and if the conditions are met, the model installs automatically, making many involuntary hosts of this local AI. To check for its presence, one can access chrome://on-device-internals/ from the address bar (it may be necessary to enable debugging URLs in chrome://chrome-urls). There, users can view the location, size, and theoretically uninstall the model. However, even if removed, Gemini Nano may be downloaded again after restarting the browser. According to Google, Gemini Nano is used for security functions that do not require sending data to the cloud, such as fraud detection or API support for developers. The trend of running AI locally has advantages in privacy and latency, but the way it has been implemented—without notice or clear user control—is controversial. Google claims that Gemini Nano is automatically removed if it detects a lack of resources and can be disabled from Settings > System by unchecking the "On-device AI" option. However, this option is not available in all installations, even with the latest version of Chrome, suggesting that the feature is not yet fully deployed or controlled. The fact that Chrome downloads and installs a 4 GB AI model without explicit consent raises questions about transparency and user control. Although Google assures that everything happens locally and that data is not shared, most users were unaware of this component's existence. It is another example of the trend to accept terms without reading them until surprises like this arise. In summary, running AI on-device makes technical sense, but the lack of communication and control from Google generates distrust. It would not be surprising if future updates expand Gemini Nano's capabilities without clearly informing the user, further diluting control over their own system.