Google's use of user data to train AI raises privacy concerns

Google has enabled a default setting that allows the company to collect users' photos, audio recordings, video files and other multimedia from services such as Search, Maps, Shopping, Flights, Hotels, Translate and News. The data can be used to train and improve its generative AI models, including Gemini.

The change was announced in a June email that introduced new controls labeled “Search History” and “Personalized Recommendations.” While the option can be turned off, users must manually disable the “Save multimedia content” sub‑setting or deactivate all web‑and‑app activity. Existing files saved before opting out may continue to be used unless they are deleted from the “My Activity” dashboard.

Google’s move reflects a broader industry trend, with Meta and OpenAI also leveraging user‑generated content for model training. Privacy advocates criticize the silent rollout and the burden placed on users to protect their data.