Google has introduced a feature that allows Gemini AI to access and use a user’s personal data from apps like Gmail, Photos, and YouTube to answer questions.
The new option, called ‘Personal Intelligence’, is in beta for paying Google AI Pro and Ultra subscribers in the US. Once enabled, Gemini can reference a user’s connected information to provide specific answers, such as retrieving a license plate number from a photo or suggesting a trip based on past emails.
Google states the feature is built with user control and privacy in mind. It is off by default, and users must manually choose which apps to connect. The company clarifies that the data from a user’s inbox or photo library is not used to train its core AI models but is accessed temporarily to fulfil a query. Users can disconnect apps or turn off the feature at any time.
Google acknowledged the beta has limitations, including potential inaccuracies and “over-personalisation”, where the AI might misinterpret a user’s interests based on their data. The company is seeking feedback from early users and plans to expand the feature to more countries and free users over time. It is not available for business or education accounts.