Google has launched Gemini Personal Intelligence in Taiwan and at the same time expanded the feature to Canada. With user permission, Gemini can access information from Gmail, Google Photos and YouTube, shifting from answers based only on general knowledge to responses tied to a person’s own context and activity across Google services.
One prompt can pull context from several Google apps
The main change is that Gemini can work across information that used to sit in separate apps. According to Google, a single request can draw on booking confirmation emails in Gmail, saved images in Google Photos and viewing history on YouTube, then combine those signals into a more practical answer.
The example in the report focuses on planning a Dragon Boat Festival trip to Penghu. Instead of checking different apps one by one, a user can ask Gemini directly and receive a full itinerary along with tailored suggestions based on relevant emails, photos and travel videos. Google said the system relies on two core capabilities: reasoning across different types of data sources and extracting detailed information from specific apps such as email and photos.
Privacy controls are built around opt-in access
Google said app connections are turned off by default. Users choose whether to enable them, and they can revoke access at any time. Once enabled, Gemini accesses data only when handling a specific task, while the underlying data remains stored in Google’s existing systems rather than being uploaded to external servers.
Gemini also shows the source of information in its responses so users can verify or correct what it used. People can save personal preferences, such as a preferred seat choice, and the system will remember them. For conversations that should not use personal context, Google said users can switch to a temporary chat mode.
Google draws a line on training data use
On model training, Google said Gemini does not directly use content from a user’s Gmail inbox or Google Photos to train the model. The company said it uses prompts entered into Gemini and the model’s replies to improve service quality.
Subscription users get access first
The rollout is happening in phases for eligible personal Google accounts. Access is being prioritized for Google AI Plus, Pro and Ultra subscribers, with free-tier users expected to gain access over the coming weeks.
To turn it on, users who receive the trial invitation in Gemini can go through “Settings” to “Personal Intelligence” and then to connected apps, where they can choose which Google services to authorize. Google said the feature works on the web, Android and iOS, and supports all Gemini models.