Google made a statement about hidden artificial intelligence downloads in Chrome

Google responded with an official statement to the allegations that the Chrome browser installs an artificial intelligence model of approximately 4 GB in size on user systems. The controversy started when Alexander Hanff, also known as The Privacy Guy, claimed that a large file named “weights.bin” was automatically downloaded with Chrome version 147. While Hanff stated that this file was one of the key components of the Gemini Nano artificial intelligence model, he claimed that the process occurred without user consent and was re-downloaded even if it was manually deleted. The issue, which quickly became a hot topic on social media and technology forums, became the focus of criticism, especially in terms of storage space and user control.

The statement from Google reveals that the file in question is not a new or secret process. Company spokesman Scott Westover told The Verge that Chrome has been using the Gemini Nano model, which has been running on the device since 2024. According to Westover, this model underlies some security functions that operate without sending data to the cloud. The model must be stored locally in order for functions such as fraud detection and certain API features available to developers to be performed on the device. Despite this, Google also states that the model is automatically removed if system resources are insufficient.

One of the focal points of the allegations was whether users could turn off this feature. Google stated that it added the “on-device artificial intelligence” option to Chrome settings as of February 2026, and users can disable the Gemini Nano model through this section. According to the company’s statement, when the feature is turned off, downloading and updating of the relevant model is completely stopped. In addition, it becomes possible to remove existing files from the system.

How does Chrome’s on-device artificial intelligence system work?

The artificial intelligence features that Google has recently placed in Chrome are largely based on the on-device approach. The company prefers to process data without sending it to the cloud, especially for privacy-sensitive security features. The Gemini Nano model is positioned as one of the fundamental parts of this approach. Thus, some analysis operations take place directly on the computer and the need to transfer user data to remote servers is reduced.

In addition, the automatic downloading of a model file of approximately 4 GB in the background has created controversy, especially for users using devices with limited storage space. However, Google argues that such models are necessary for modern AI functions. The company’s statement emphasizes that the model is only activated on supported systems and is managed according to device performance. Still, some users think that such features, which are enabled by default, should be announced more visibly.

Users who want to disable the Gemini Nano model in Chrome must turn off the “on-device artificial intelligence” section in the browser settings. After this process, new versions of the model are not downloaded and background updates are stopped. Although Google’s statement rejects the claims that the system in question works completely secretly, it shows that the discussions about storage usage and user control of artificial intelligence models running on the device will continue.