Google Confirms Its New AI Cursor Sends Your Screen To Gemini

Google has confirmed that its new AI-powered cursor, called Magic Pointer, sends your screen content to its Gemini AI to suggest actions — raising fresh privacy alarms just as regulators tighten scrutiny on tech giants. The feature, rolling out on Google’s Chromebook laptops, analyzes whatever is on screen in real time, from emails to web pages, and offers contextual suggestions like “compose a reply” or “summarize this document.” This means every pixel you see is potentially transmitted to Google’s cloud servers for AI processing. Google insists the data is anonymized and not stored, but privacy advocates warn the practice could erode user trust and invite regulatory action under laws like Europe’s GDPR and upcoming AI acts. The announcement comes months after Microsoft’s similar Recall feature was delayed over privacy backlash. Magic Pointer is part of Google’s broader push to embed Gemini across all its products, integrating deep AI assistance into the operating system itself. The feature is opt-in by default, but critics argue casual users may not understand the implications. Google has not disclosed whether screen data is used to train AI models, only that it is processed transiently. The company also faces questions about third-party app data: if a user has a banking app or confidential document open, that text could be seen by Gemini. Google says the cursor only activates on selected screens and users can disable it entirely in settings. For now, Magic Pointer is exclusive to newer Chromebook models, but expansion to other Google laptops and possibly desktops is expected. The broader implication is that AI copilots are moving from being optional assistants to always-on observers of our digital lives. What happens next: privacy watchdogs may issue guidelines or launch investigations; Google may publish a transparency report; and users will vote with their clicks — or their clicks will be watched.