Google Search update stores user media uploads for AI model training
Tags AI · Consumer
Google updated its Search history feature to store media uploads from user interactions, including images used in reverse image searches, for training its AI models. The change was reported by WIRED and generated significant privacy concerns. Users can opt out of the new data retention feature through their Google account settings. The update reflects Google's broader strategy of leveraging user interaction data to improve its AI models, but raises questions about informed consent when the data use case extends beyond the original interaction purpose.
Technical significance
The boundary between user data and AI training data continues to blur. Companies that rely on user interaction data for model training face increasing regulatory and reputational risk. The technical challenge is building training pipelines that can effectively use interaction data while respecting granular consent preferences — a non-trivial engineering problem at Google's scale.