Google clarifies how Gemini in Gmail handles personal email

Google published a privacy clarification on 7 April saying it does not use personal emails to train its foundational AI models, including Gemini. It described access granted to Gemini inside Gmail as limited to individual tasks, such as summarising an email.

The company also said Gemini in Gmail processes information to complete the requested task and does not retain that data afterward. These are Google’s stated practices for the Gmail integration. They answer two different concerns: whether inbox content becomes training material, and whether the assistant keeps the content after a request. The statement concerns this particular product experience; privacy terms for another Google AI service need to be checked separately.