Google I/O 2026 kicked off today with a keynote presentation that reaffirmed the company's commitment to artificial intelligence across its entire ecosystem. From a new generation of Gemini models to a redesigned Search experience and early glimpses at Android XR smart glasses, the event showcased how Google plans to embed AI deeper into everyday life. Here is a comprehensive look at all the major announcements.
Gemini Gets a Major Overhaul
Gemini remained the star of the show. Google announced a new family of models called Gemini 3.5, with the 3.5 Flash model delivering faster and more intelligent responses. The company also introduced Gemini Omni, a multimodal model family capable of generating videos from text, images, audio, and video inputs. The first model, Omni Flash, is positioned as a video-generation counterpart to the popular Nano Banana image model.
The Gemini app itself received a visual redesign dubbed Neural Expressive, featuring fluid animations, vibrant gradients, and a simplified interface centered around an 'Ask Gemini' bar. Haptic feedback and new typography complete the refresh. Alongside the redesign, Google introduced Gemini Spark, an always-on AI agent that runs 24/7 on Google Cloud virtual machines. It can autonomously write emails, create study guides, monitor subscriptions, and connect with third-party apps via the Model Context Protocol. Integrations with Canva, OpenTable, and Instacart were announced at launch.
Gemini Spark is powered by Gemini 3.5 Flash and is designed to handle persistent tasks in the background. Users can leverage it for everything from tracking package deliveries to generating customized to-do lists based on email content. This represents Google's answer to the growing trend of AI agents that proactively assist users without requiring constant prompts.
Search: The Biggest Changes Ever
Google Search received what the company described as its most significant update. The search box now dynamically expands as users type longer queries, and AI-powered suggestions go beyond traditional autocomplete. AI Overviews, the AI-generated summaries at the top of search results, become more reliable when asking natural-language questions. If users want to follow up, they are seamlessly redirected to AI Mode, a chatbot-like interface that accepts attached documents, photos, videos, and Chrome tabs.
Robby Stein, Google's vice president of product for Search, confirmed that AI Overviews will appear 'reliably' for natural-language queries. The underlying engine is the new Gemini 3.5 Flash model, which also powers the expanded capabilities of AI Mode. This update blurs the line between traditional search and conversational AI, making it easier to conduct complex research or ask multi-step questions directly from the search bar.
Gmail Goes Live and Gets Smarter
Gmail introduced Gmail Live, a voice-activated AI assistant that lets users interact with their inbox by speaking. By tapping a new icon in the search bar, users can ask questions about events, travel details, or specific emails, and Gmail Live retrieves relevant information from the user's inbox and displays it in a dedicated interface. During a demo, a Google employee asked about a school event and an upcoming trip, and the system quickly pulled dates, locations, and confirmation details.
Additionally, Gmail's AI inbox features expanded. The service can now generate custom to-do lists and draft personalized replies based on email content. These features leverage Gemini's deep integration with Google Workspace, allowing the AI to understand context across Docs, Sheets, and Calendar. The always-on nature of Gemini Spark further extends Gmail's capabilities, enabling automatic monitoring for subscription fees or important deadlines.
Wear OS 7 Brings Live Updates and Widgets
Wear OS 7, the next major version of Google's smartwatch platform, was announced with several consumer-friendly features. The most notable addition is Live Updates, borrowed from Android's iPhone-style live activities, which can display real-time information for deliveries, sports scores, and even ongoing AI tasks directly on the wrist. Wear Widgets, a new card-like interface, offers glanceable information in 2x1 or 2x2 layouts that align with Android's widget system.
Google also introduced Gemini Intelligence for select watches launching later this year. This branding covers personalized and proactive AI features, such as smart replies and contextual recommendations based on user activity. The update positions Wear OS as a more capable companion for AI-driven daily use.
Android XR Smart Glasses Take Shape
After months of rumors, Google provided concrete details about its Android XR smart glasses, under the project name Project Aura. Developed in partnership with Xreal, the glasses are dark sunglasses with a built-in heads-up display and a separate compute puck. The puck now includes a fingerprint scanner for secure unlocking and can be worn on a lanyard or clipped to a waistband. Google revealed that multiple fashion collaborations are in the works, including models from Warby Parker and Gentle Monster, and hinted at designs from Samsung and Gucci.
During a hands-on demo, journalists experienced the glasses' ability to overlay navigation, translations, and notifications. The device runs on Android XR, a specialized OS for augmented reality. Google expects the first consumer models to launch before the end of the year, with pricing yet to be announced.
Shopping and Commerce: Universal Cart and AI Agents
Google is doubling down on AI-driven shopping with a Universal Cart that works across retailers and Google products. Users can add items while browsing Search, chatting with Gemini, or watching YouTube videos (and later Gmail), and then check out directly through Google. The cart tracks prices, sends in-stock alerts, suggests discounts, and warns about potential issues. This move positions Google as a central commerce hub, integrating AI agent capabilities to automate purchases.
The company also detailed how Gemini Spark can handle financial tasks, such as monitoring credit card statements for hidden fees or recurring charges. These agentic features require user permission but represent a significant step toward letting AI manage personal finances.
AI Detection and Security
In response to growing concerns about deepfakes, Google expanded its AI detection tools. SynthID watermarks, developed by DeepMind, are now integrated into Chrome and Search. Users can right-click (or circle) an image and ask 'Is this AI generated?' to check for SynthID markers or C2PA content credentials. This multi-layered approach aims to make provenance information more durable across platforms.
OpenAI also announced it would adopt SynthID watermarks alongside its existing C2PA commitments, indicating industry-wide cooperation on AI labeling. Additionally, Google opened the API for CodeMender, an AI agent for code security, to select experts. CodeMender can flag and fix vulnerabilities automatically, competing with tools like Anthropic's Mythos. Google sees cybersecurity as a key revenue driver for its AI platform.
Pricing and Subscription Updates
Google revised its AI Ultra plan pricing. The top-tier subscription, previously $249.99 per month, now starts at $100 per month. A $200 tier includes access to Project Genie, Google's world model that generates interactive AI experiences grounded in real-world Street View data. This aligns with OpenAI's own $100 and $200 Pro tiers, reflecting a broader industry shift toward more accessible high-end AI subscriptions.
Other Notable Announcements
Google introduced Pics, an AI image editing app for Workspace that allows users to click on any part of an image and leave a note describing a desired change, much like commenting on a Google Doc. Powered by Gemini and Nano Banana 2, Pics simplifies iterative image editing.
AI Studio now supports building native Android apps entirely through natural language prompts. Users can describe an app idea, preview it in an embedded emulator, and install it directly on a connected device. This 'vibe coding' approach lowers barriers to mobile development.
Volvo announced that its upcoming EX60 SUV will integrate Gemini with external cameras, enabling the AI to interpret parking signs, lane markings, and nearby points of interest. This marks a practical application of generative AI in automotive contexts.
Google DeepMind CEO Demis Hassabis closed the keynote by suggesting we may be entering the 'foothills of the singularity,' referencing the accelerating capabilities of AI. He introduced Gemini for Science, a set of tools aimed at accelerating drug discovery and other scientific research.
Pointer remote support was announced for Google TV, allowing developers to add motion-controlled navigation on home screens and within apps. This could lead to new hardware remotes similar to LG's magic remote but native to the Android TV ecosystem.
Finally, Google's Project Genie world model gained the ability to generate interactive experiences grounded in real-world locations using Street View data, enabling users to explore AI-generated versions of actual places.
Source: The Verge News