Google announced Thursday new artificial intelligence features for Gmail designed to personalize users’ writing, answer questions drawn from inbox content, and generate daily to-do lists — an expansion the company said is aimed at turning the world’s most popular email service, which counts more than 3 billion users, into a proactive personal assistant.
The additions, powered by Google’s Gemini 3 model, could transform how users interact with their inboxes, though they also renew privacy questions about how deeply the company’s AI systems will access personal correspondence.
Three new tools
The most broadly available of the new features is “Help Me Write,” which learns individual users’ writing styles to personalize outgoing emails and offer real-time drafting suggestions.
Subscribers to Google’s paid Pro and Ultra tiers will gain access to a Gmail search upgrade modeled on the AI Overviews that Google has built into its main search engine. The tool allows users to pose conversational questions in Gmail’s search bar and receive instant answers drawn from their inbox contents.
A third feature, called “AI Inbox,” is being rolled out to a subset of “trusted testers” in the United States. When activated, it sifts through a user’s inbox and suggests to-do lists and topics the user might want to explore.
“This is us delivering on Gmail proactively having your back,” said Blake Barnes, a Google vice president of product.
The new tools will initially be available only in English within the United States. Google said it plans to expand them to other countries and languages as 2026 progresses.
Powered by Gemini 3
All three features are tied to Gemini 3, the AI model Google added to its search engine in late 2025. That upgrade drew notice across the technology industry: Sam Altman, the chief executive of OpenAI — whose company makes the ChatGPT chatbot — issued a “code red” following Gemini 3’s arrival in search, according to the Associated Press.
Gmail itself has amassed its user base over roughly 22 years, growing since its introduction around 2004 to become nearly as widely used as Google’s search engine.
Privacy questions
Allowing AI to read more deeply into personal inboxes renews privacy concerns that have followed Gmail since its early years. To subsidize the free service, Google originally embedded targeted ads based on email content, a practice that drew criticism from lawmakers and consumer advocacy groups before becoming common across major email providers.
Google said the content that its AI analyzes within Gmail will not be used to train Gemini models. The company said it also built an “engineering privacy” barrier to contain inbox information and protect it from external access.
Users can proofread AI-generated messages before sending them and can disable the features at any time, Google said.