How to Disable Google's Gemini in Chrome

· Source: WIRED - Ai · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Cybersecurity & Data Privacy · Depth: Novice, long

Summary

Google's Chrome browser for desktop is silently installing a 4 GB Gemini Nano AI model on users' computers, specifically on macOS and Windows systems, without explicit consent. This on-device model, which occupies approximately 4.27 GB, is found in a hidden folder within the Chrome application directory. While the model is designed for local AI processing and does not send data to Google's servers, its unconsented installation raises concerns about user autonomy, system bloat, and resource consumption. Users can disable this "optimization" by navigating to `chrome://flags/#optimization-guide-on-device-model` in their browser and restarting Chrome. The issue highlights a broader critique of Big Tech's approach to installing software and features without clear user permission, contrasting with more opt-in approaches seen in other platforms like Apple Intelligence.

Key takeaway

For engineering leaders and privacy-conscious users, Google's unconsented installation of the Gemini Nano AI model in Chrome underscores the importance of auditing software dependencies and default configurations. You should evaluate your organization's browser strategy, potentially migrating to privacy-focused alternatives like Brave, Firefox, or Waterfox, to ensure greater control over system resources and data. This incident highlights a need for explicit opt-in mechanisms for significant software installations.

Key insights

Google Chrome installs a 4GB Gemini Nano AI model without user consent, raising privacy and system bloat concerns.

Principles

Method

To disable the Gemini Nano AI model in Chrome, navigate to `chrome://flags/#optimization-guide-on-device-model` in the URL bar, then select "Disabled" from the dropdown menu and restart the browser.

In practice

Topics

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Software Engineer, IT Professional, General Interest

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by WIRED - Ai.