Gemini Adds Notebooks — Privacy Pinch

Google is turning Gemini into a filing cabinet, not just a chatbot. The new “notebooks” feature lets people group chats, files, websites, and instructions into one project space inside Gemini, and Google started rolling it out on April 8 for Google Artificial Intelligence Ultra, Pro, and Plus subscribers on the web. (blog.google) A notebook is basically a folder with a memory. You can name it, pull in old Gemini conversations, add documents from your computer or Google Drive, paste in text, and attach web links so the assistant answers from that pile instead of from a blank page. (pcmag.com) Google tied the feature directly to NotebookLM, which is the company’s research tool for working from source material. If you add a source in Gemini, it shows up in NotebookLM too, and Google says that lets you jump into NotebookLM features like Video Overviews and Infographics without rebuilding the project from scratch. (blog.google) This is Google chasing a more practical use for artificial intelligence at work: long projects. A one-off chatbot answer is useful for 30 seconds, but a persistent project space is what teams need when they are tracking a product launch, a legal brief, a grant application, or a semester of class notes. (blog.google) Google is also widening the audience around the same time. The company has been pushing Gemini deeper into Google Workspace, and recent updates have focused on more language coverage and more places where Gemini can pull context from the tools people already use every day. (workspaceupdates.googleblog.com) (blog.google) That convenience is where the privacy tension starts. In March, Google said Gemini in Docs, Sheets, Slides, and Drive can pull relevant information from files, emails, and the web when users choose sources, which makes the assistant more useful but also means the system can touch much more of a person’s work history. (blog.google) Gmail is the most sensitive example because an inbox is part to-do list, part archive, and part confession booth. Google said on April 7 that Gemini in Gmail handles “isolated tasks” like summarizing a long thread, does not train Gemini’s foundational models on personal emails, and does not retain inbox data after completing the request. (blog.google) Google makes a similar promise for paid Workspace use. Its help pages say Gemini in Gmail, Calendar, Chat, Docs, Drive, Sheets, Slides, Meet, and Vids uses Workspace content to answer prompts, but does not use that content to train or improve Gemini or other generative artificial intelligence models. (support.google.com) The catch is that Google’s protections are split across products and account types. The same help page says if a user with a personal account chooses to share Workspace data with Gemini Apps or Search for “Personal Intelligence” or through screen actions like screenshots, that data is handled under those separate terms and may be used for model training and improvement. (support.google.com) So the real story is not just “Google added folders to Gemini.” Google is trying to make Gemini useful enough to sit inside the places where people keep their most valuable information, and every step closer to Gmail and Workspace makes the product better at work while making the privacy promises more important to read line by line. (blog.google 1) (blog.google 2)