Google stuffs Chrome full of AI features whether you like it or not (www.theregister.com)

Google is baking its Gemini AI across Chrome starting today for US macOS and Windows users with English set — adding a persistent Gemini button to pages and an “AI Mode” button in the omnibox. The new tools let Gemini summarize any page, synthesize info across multiple tabs, search your web history and linked Google apps, and perform chatbot-style queries from the toolbar. Google clarified that pressing Enter in the omnibox still triggers regular search; querying AI Mode requires a new on-screen search button. An upcoming wave of agentic features will let Gemini act on webpages for you (booking, ordering, etc.), automate password changes on supported sites (listed examples include Coursera, Spotify, Duolingo, H&M), and try to detect scams or notification-spam using learned web preferences. This is consequential because it shifts Chrome from a neutral browser into an AI-first interface, re-accelerating “AI browser” competition and raising product‑strategy and monetization stakes (and potential traffic loss for creators). Technical and privacy implications are significant: agentic web actions require new permission models, automated password workflow needs site support, and it’s unclear whether browsing data will be used to train models. Google hasn’t confirmed opt-in defaults or training-use policies, so users and publishers should watch rollout controls, permissions dialogs, and data‑use disclosures closely.