Google Chrome Installed a 4GB AI Model on Your PC (oztalking.com)

Google Chrome has installed a 4GB on-device AI model named Gemini Nano without user consent, leading to significant privacy concerns and storage issues for users. The installation process checks the hardware specifications of the user's device and downloads the model in the background, automatically reinstating it if deleted. This controversial move was revealed by Swedish privacy researcher Alexander Hanff, leading to discussions on the implications of tech companies leveraging user devices as part of their AI infrastructure without explicit permission. The significance for the AI/ML community lies in the broader question about user consent and privacy as AI moves from cloud-based solutions to on-device processing. While on-device AI can enhance features such as scam detection and writing assistance—argued by Google to protect user privacy—it paradoxically violates privacy by not allowing users to control installation. Critics contend this practice contradicts existing privacy regulations, like the EU’s ePrivacy Directive. As companies increasingly integrate AI functionalities into users' devices, the need for a redefined consent framework becomes urgent, raising ethical questions about ownership and control over personal technology resources.