4 min readUpdated: May 8, 2026 09:16 AM IST
Google’s Chrome browser is facing scrutiny after a security researcher claimed that the browser quietly downloads and stores a large on-device AI model without clearly asking users for permission.
The issue centres around Gemini Nano, Google’s lightweight artificial intelligence model designed to run directly on a device instead of relying entirely on cloud servers. According to researcher Alexander Hanff, also known online as “That Privacy Guy”, Chrome may automatically install nearly 4GB of AI-related files on compatible Windows and macOS systems.
The discovery has sparked concerns about privacy, storage usage, and transparency, especially since many users didn’t know the files were being stored on their devices.
Hanff claims the AI model is located inside a folder named “OptGuideOnDeviceModel,” with a large file called “weights.bin” believed to contain the Gemini Nano model weights. According to the researcher, the files are automatically downloaded on systems that meet Chrome’s hardware requirements, even if users are not actively using Gemini features.
The researcher further said that manually deleting the files does not permanently remove them. In some cases, Chrome reportedly downloads the AI model again after the browser restarts. Several users across online forums and tech communities have also discussed finding the files on their systems over the past year.
Google has responded to the concerns by confirming that Gemini Nano has been part of Chrome since 2024. The company says the model powers several AI and security-related features, including scam detection and developer APIs, while reducing the need to send certain data to cloud-based servers.
According to Google, the AI model may automatically uninstall itself if a device is running low on resources. The company also stated that it has started rolling out options allowing users to disable and remove the feature directly from Chrome settings.
Story continues below this ad
Gemini Nano is Google’s smaller AI model built specifically for on-device processing. Unlike larger cloud-based AI systems, Nano runs locally on a user’s computer, enabling faster responses and potentially improving privacy for certain tasks.
Google has been rapidly expanding Gemini AI across its ecosystem, including Android smartphones, Search, Workspace apps, and Chrome. In the browser, the AI model is reportedly used for tasks such as scam detection, AI-assisted browsing features, and developer tools.
Here’s a way to check
Users who want to check whether the AI model is installed can type “chrome://on-device-internals” into Chrome’s address bar. This page reportedly shows whether Gemini Nano is present and how much storage space it occupies.
If available, users can disable the feature through Chrome’s settings menu under the system section. Certain experienced users have also recommended manually disabling Chrome flags relating to AI, but this may impact the browser’s performance or its AI-based capabilities.
Story continues below this ad
The issue indicates the wider controversy concerning the use of AI in consumer technology products. With technological firms integrating artificial intelligence into operating systems, web browsers, and other consumer devices, privacy advocates insist that any large-scale AI model should not be automatically installed without a user’s informed consent.
Despite Google’s assertion that Gemini Nano enhances user security and creates engaging artificial intelligence experiences, the issue points to increasing societal concerns about the hidden integration of AI in popular applications.
© IE Online Media Services Pvt Ltd
