Google has launched a new feature that allows its Gemini artificial intelligence to use personal data from apps such as Gmail, Photos, and YouTube to answer user questions, the company said.

The option, called Personal Intelligence, is in beta and available to paying Google AI Pro and Ultra subscribers in the United States. Once enabled, Gemini can draw on connected data to deliver tailored responses, such as finding information in emails or identifying details in photos.

Google said the feature is designed with privacy controls and is turned off by default. Users must choose which apps to connect and can disconnect them or disable the feature at any time. The company said that personal data accessed during queries is not used to train its core AI models and is used only temporarily to provide answers.

Google acknowledged the beta version has limits, including possible inaccuracies and over-personalisation, where the system may misread a user’s interests. It said feedback from early users will guide further development.

The company plans to expand the feature to more countries and eventually to free users. It is not available for business or education accounts, Google said.