Google is connecting its Gemini artificial intelligence chatbot to users' personal photo libraries, allowing the company's image generator to create pictures featuring real people drawn from their own photographs.
The update, announced on Thursday, means that someone who has labelled family members and pets in Google Photos can ask Gemini to generate an image such as "create a claymation picture of me and my family enjoying our favourite activity" and receive a result that actually resembles the people in their life.
Previously, getting that kind of personalised output required writing long, detailed descriptions and manually uploading reference photographs, a process that most people found cumbersome enough to abandon.
The feature is powered by Nano Banana 2, Google's image generation system built into Gemini, and works through a broader framework the company calls Personal Intelligence, which was launched in January and allows Gemini to draw on data from across a user's Google account, including Gmail, Calendar, Drive, YouTube viewing history and search activity.
Personal Intelligence is switched off by default and must be deliberately enabled, with users choosing which individual applications Gemini is allowed to access.
When the photo connection is active, Gemini uses the labels that Google Photos already assigns to faces, relying on names and group descriptions such as "family" or "Dad" to understand who a user is referring to in a prompt.
Google has built in a transparency mechanism: after each image is generated, users can tap a sources button to see which photographs from their library were used as reference material.
If the system selects the wrong image, users can tap a plus icon to substitute a different photograph from their library.
The company said Gemini does not directly train its AI models on users' private photo libraries, a distinction likely designed to address the privacy concerns that inevitably accompany any feature asking people to hand over personal images to an AI system.
Google did acknowledge, however, that it uses the text of prompts and the AI's responses to improve its products over time, a broader data use that applies across Gemini regardless of whether the photo feature is enabled.
The structural advantage Google holds in this area is difficult for rivals to replicate: no other company has the same depth of personal data across email, photographs, calendars, search history and video consumption already sitting in a single account, ready to be connected.
Related reading
- Nvidia quantum AI launch triggers buying frenzy across Asian and US tech stocks
- Google cracks down on 'back button hijacking' in Chrome
- Google opens its music generation model to developers, with watermarking built in from the start
The feature is rolling out over the coming days to paying subscribers in the United States on Google's AI Plus, Pro and Ultra plans, with free users and international markets expected to follow.
Europe is notably absent from the initial launch, likely reflecting the stricter data protection rules under the European Union's General Data Protection Regulation that complicate any feature combining personal data with AI processing.
The recap
- Google links Gemini image generator to users' Google Photos library.
- Personal intelligence is available only on paid Google AI plans.
- Users can review source images or manually select photos.
