Gemini Now Infuses Personal Photo Context Into AI Image Generation

When you prompt the chatbot to design a dream house or select desert island essentials, the system scans your connected Google Photos account to identify recurring themes, friends, and family members. According to Google spokesperson Elijah Lawal, the integration relies on existing photo labels to ground the generated output in your reality rather than generic stock imagery.

The company maintains that opting into this feature does not grant permission to directly train its core AI models on your private media library. Google clarifies that while it may analyze specific prompts and subsequent model responses for limited training purposes, your raw photo data remains outside the primary training loop. This update rolls out over the coming days for AI Plus, Pro, and Ultra subscribers in the United States, with a broader expansion to Chrome desktops and additional user bases expected shortly.