# Google Harvests Your Data to Train Gemini AI—Here's How to Opt Out > Source: > Published: 2026-06-26 13:31:02+00:00 Google is quietly expanding its data harvesting operation to feed its Gemini AI models—pulling from your searches, your maps, your translations, and your uploaded photos—and most Americans have no idea it's happening. The stakes are straightforward: every search you make, every photo you upload for assistance, every language you practice on Google Translate is now fuel for an AI system controlled by one of the most powerful corporations on Earth. These systems will shape what information you see, what gets recommended to you, and what gets buried. The surveillance capitalism model that built Google's advertising empire is now being weaponized to train the next generation of AI—and opting out requires navigating a maze the company has no incentive to make obvious. According to BGR, Google is rolling out new settings governing data collection from Google Search, Google Lens, Google Maps, Google Translate, Google News, as well as Shopping, Flights, and Hotels searches. The data swept up includes "your searches, info from sites you visit with Search services, and generative AI responses," along with general location data and uploaded media—photos, videos, audio, and files. Interactions with Google Lens, Search Live, and Google Translate, including foreign language speaking practice, are all fair game. A new "Search Services History" menu will eventually appear in user accounts containing the controls to disable AI training. The rollout will take months, Google says. Until then, users can manage existing controls through the My Activity page in their Google Account settings. Once the new menu appears, turning off Search Services History will prevent your data from being used to train future Gemini models. Google's assurances are the usual Silicon Valley comfort blanket. The company claims data will be "disconnected from your Google account" before human review and AI training. It says filters remove "a broad range of identifying info or sensitive personal information." It promises to ask permission before sharing media content for human review. Education institution accounts, notably, are excluded from AI training by default—suggesting Google knows exactly how problematic this data grab is when users are paying attention. Meanwhile, your government is making the problem worse. WTOP reports that the District of Columbia's new DC.gov website—built at a cost of $500,000—will include an AI-powered search tool built using Google's AI technology. DC Chief Technology Officer Stephen Miller said the city is "trying to clean up our content" to ensure the AI gives correct information, and envisions a future where residents can say "Hey Siri, renew my driver's license." So now citizens interacting with basic government services will be routed through Google's AI infrastructure—feeding the same data machine they should be opting out of. The picture is clear: Google expands its data collection across the apps Americans use daily, then embeds its AI into government portals so even civic interactions flow through its systems. The opt-out exists, but only if you know to look for it. **How to opt out now:** Go to your Google Account's My Activity page and disable search history tracking. When the new Search Services History menu appears in your account in the coming months, disable it immediately. If you use Gemini directly, check its settings to opt out of chat data being used for training. The controls exist—Google just isn't going out of its way to highlight them. The question isn't whether Google will honor its opt-out promises today. It's whether a company with this much power over information flow will honor them tomorrow, when the regulatory pressure is off and the AI models are already built.