Chrome's Silent 4GB AI Install Sparks Privacy Fears

Between late April and early May 2026, Google Chrome desktops began silently installing a 4GB AI model named Gemini Nano. This automatic rollout, identified by security researcher Alexander Hanff, proceeded without any user consent prompts or straightforward opt-out options. This stealthy installation has raised significant privacy concerns, with some critics suggesting it may violate European data protection laws.

Gemini Nano is designed for on-device functions like scam call detection and text message assistance, differentiating it from cloud-based AI queries. Google stated that the model uninstalls if a device lacks sufficient resources. They also mentioned that users can now disable and remove the model directly within Chrome settings. However, Hanff posits that Google's decision to install the AI without explicit permission could be an effort to reduce their own operational costs, moving AI processing to user hardware.

Concerns also extend to potential legal ramifications, particularly concerning the EU's General Data Protection Regulation's principles of lawfulness, fairness, and transparency. Hanff noted Google's extensive history of privacy issues, suggesting that bypassing explicit permission was a strategic choice to facilitate the deployment of this AI model and subsequent technologies. Removing Gemini Nano involves accessing Chrome settings, disabling the 'On-device AI' toggle, and potentially deleting specific local folders.