Chrome quietly downloads a 4 GB on-device Gemini Nano model (weights.bin) into a per-user profile via the component update system, without user consent or an opt-out option.

Independent analyst and lawyer Alexander Hanff argues the installation happens by default when AI features are active, with no obvious opt-in prompt or explicit consent.

The download occurs across billions of devices through Chrome's update mechanisms, triggering backlash over consent, disclosure, and transparency.

Environmental impact is highlighted, citing electricity use and storage-related emissions comparable to tens of thousands of homes per year.

Legal analysis claims breaches of EU ePrivacy Directive Article 5(3), GDPR principles (lawfulness, fairness, transparency), and GDPR data-protection-by-design considerations, with UK GDPR and CCPA noted.

Public commitments lack transparent accounting of energy and carbon trade-offs between on-device versus cloud inference, calling for more rigorous analysis.

Chrome’s immense popularity could amplify environmental and legal consequences, turning this into a test of responsible AI and sustainability commitments by major tech platforms.

Bottom line: this is a transparency and storage-management concern as AI features migrate to on-device processing, not evidence of malicious intent.

The deployment creates opportunities and constraints for startups: a potential market for opt-in local inference tools, but a disadvantage for competing models already installed on devices.

Practical remedies include registry edits on Windows and Chrome flag adjustments on Mac to disable or reclaim control over the on-device model.

Suggested actions for Google: obtain explicit consent, offer opt-in or trigger-based downloads, surface and allow removal of downloaded models, publicly document behavior, respect deletions, disclose aggregate emissions, and provide retrospective consent notifications.

Backlash signals a user preference for consent, transparency, and control, potentially shaping future disclosure and management of AI features by browser vendors.