Gemini adds crisis help and shopping tools

Google is teaching Gemini to do two very different jobs at once: steer people in crisis toward a real human faster, and steer shoppers toward a product faster. The new update adds one-tap hotline access for mental-health emergencies and new shopping features across the Gemini app, Search artificial intelligence mode, and Circle to Search. (blog.google) (engadget.com) (blog.google) The crisis side comes first because chatbots keep getting used for moments they were never supposed to handle alone. Google said more than 1 billion people worldwide are affected by mental-health conditions, and it said users increasingly turn to artificial intelligence tools during vulnerable moments. (blog.google) A chatbot can generate words instantly, but a hotline can connect someone to a trained person who can act in real time. Google’s redesign is built around that gap: instead of only showing information, Gemini now surfaces a crisis hotline module with a one-touch interface that lets distressed users reach outside help more quickly. (engadget.com) (blog.google) Google also changed how Gemini writes back when a message suggests a severe mental-health crisis. The company said the system now aims to respond in a more careful way and guide people toward professional support rather than treating the exchange like an ordinary chat. (blog.google) (9to5google.com) This update did not appear in a vacuum. Engadget reported that Google’s changes arrive as artificial-intelligence companies face growing scrutiny over whether conversational systems can worsen harm when users are isolated, delusional, or suicidal. (engadget.com) (blog.google) Google says its mental-health work is based on research and clinical best practices, and it says it consulted medical and mental-health professionals on Gemini’s safeguards. That language matters because companies are trying to draw a line between an assistant that can offer information and a system that should never pretend to replace care. (blog.google 1) (blog.google 2) Then there is the other half of the announcement, which is about shopping rather than safety. Google rolled out three Gemini-powered shopping features in India across the Gemini app, Search artificial intelligence mode, and Circle to Search, using Google’s Shopping Graph, which the company says contains more than 50 billion product listings. (blog.google) (moneycontrol.com) The first feature puts shopping directly inside the Gemini app. Instead of typing a few keywords and opening ten tabs, users can describe what they want in plain language and get product suggestions, prices, comparisons, reviews, and buying links inside the conversation. (blog.google) (moneycontrol.com) The second feature extends that same style into Search artificial intelligence mode. Google says users can ask more layered shopping questions there, and the system can narrow choices using the Shopping Graph’s product data instead of making people reformulate the same search over and over. (blog.google) (gadgets360.com) The third feature changes Circle to Search from a tool that identifies one obvious item into one that can help with several items in the same image. Google says people can now circle or tap objects on screen and get shopping results tied to what they are looking at, which turns a screenshot or photo into something closer to a storefront. (blog.google) (moneycontrol.com) The geography matters here. Google’s shopping rollout was announced for India, where the company has been layering new commerce tools onto Search and where it previously launched Virtual Try-On for apparel listings. (blog.google) (gadgets360.com) Put together, the two announcements show Google pushing Gemini in opposite directions at the same time. In one direction, the company is trying to make the system step back and hand off urgent human problems to real-world support; in the other, it is trying to make the system step in more deeply as a buying assistant for everyday decisions. (blog.google 1) (blog.google 2) That split may end up being the clearest picture of where big consumer artificial intelligence products are heading in 2026. The same assistant is being trained to recognize when it should do less, and to become useful enough in lower-stakes tasks that people keep it open for more of the day. (engadget.com) (blog.google)