Google said it is adding new artificial intelligence features to Gmail, aiming to improve writing, summarize information found in inboxes, and deliver daily to-do lists. The company announced the changes Thursday, with initial availability in English for users in the United States.
Google said the new Gmail AI options will expand beyond the initial U.S. English rollout as the year unfolds, including other countries and other languages. One of the most broadly available options will be a “Help Me Write” feature that is designed to learn a user’s writing style and make real-time suggestions on how to improve emails.
Google also said subscribers who pay for its Pro and Ultra services will be able to use technology in Gmail that mirrors the AI Overviews built into Google search. The expansion, Google said, will let those subscribers pose conversational questions in Gmail’s search bar to get instant answers about information they are trying to retrieve from their inboxes.
In another feature preview, Google said “AI Inbox” is being rolled out to a subset of “trusted testers” in the United States. When the feature is turned on, Google said it will sift through inboxes and suggest to-do lists and topics users might want to explore.
“This is us delivering on Gmail proactively having your back,” said Blake Barnes, a Google vice president of product.
Google said all of the new Gmail AI technology is tied to its latest AI model, Gemini 3. The AP reported that Gemini 3 was unleashed into Google search late last year, and that Google designed the upgrade to turn Google search into a “thought partner.”
The AP said the Gemini 3 upgrade was so well received it prompted OpenAI CEO Sam Altman to issue a “code red” after its release. The AP also noted that pushing more AI into Gmail poses potential risks—particularly if the technology malfunctions and presents misleading information or crafts emails that get users into trouble—though it said users can proofread messages or turn off the features at any time.
The AP raised privacy concerns as another challenge. It said allowing Google’s AI to dig deeper into inboxes to learn more about users’ habits and interests could raise privacy issues, and that Gmail has faced such concerns before; it described an early privacy backlash over targeted ads in Gmail based on information contained within electronic conversations.
Google said it will not use any content that the AI analyzes to train the models that help Gemini improve. The AP reported that Google also built an “engineering privacy” barrier intended to corral the information within inboxes to protect it from “prying eyes.”
The AP said Gmail, introduced nearly 22 years ago, has amassed more than 3 billion users, making it nearly as ubiquitous as Google’s search engine. It said Gmail’s new AI options could represent another pivotal moment in the evolution of the email service.