MOUNTAIN VIEW, Calif. – At its annual I/O developer conference held near its Silicon Valley headquarters on Tuesday, Google laid out an ambitious new vision to transform its iconic search bar into a proactive artificial intelligence assistant capable of handling end-to-end user tasks, from restaurant reservations to personalized news alerts and direct outreach to businesses, all triggered by a simple natural language request.
The announcement marks a major milestone in Google’s three-year sprint to catch up to OpenAI’s ChatGPT, the generative AI breakthrough that upended the company’s decades-long dominance in online search and forced a rapid company-wide pivot to AI integration. Early growth metrics signal significant traction for Google’s existing AI offerings: the flagship Gemini AI app now boasts 900 million monthly active users, double its user base from just 12 months ago, while the company’s AI-powered search overhaul, AI Mode, has already hit one billion monthly users globally.
Opening the event, Google CEO Sundar Pichai introduced the company’s newest flagship AI product: Gemini Spark, a dedicated personal AI agent that will roll out to premium Google One subscribers in the U.S. starting next week. “Search is evolving beyond isolated, one-off queries into a continuous, contextual conversation that unlocks deeper insights and connects users to the full breadth of information on the web,” Pichai told reporters on the sidelines of the conference. “That’s the future we’re building with this new generation of tools.”
A broader upgrade to Google’s core search engine will follow this summer for U.S. users, introducing always-on AI agents that can proactively notify users of breaking news, complete booking transactions, and communicate with third-party businesses on a user’s behalf. These new features tap into the fast-growing “agentic AI” trend that has swept Silicon Valley since late 2025, when Austrian developer Peter Steinberger launched OpenClaw, a pioneering platform that enabled AI to complete complex multi-step tasks ranging from booking flights to sorting email inboxes and building custom applications from simple chat prompts. OpenAI poached Steinberger earlier this year, setting off a fierce race among the world’s largest tech companies to bring agentic AI capabilities to mainstream consumers – even as experts raise persistent concerns about security risks and the exorbitant computing costs required to run these sophisticated tools.
To maintain an edge over rivals OpenAI and Anthropic, Google also launched Gemini 3.5 Flash, its newest lightweight large language model, on Tuesday. The company claims the model runs four times faster than top competing models including Anthropic’s Claude Opus and OpenAI’s ChatGPT 5.5, while matching their performance on core AI tasks. Gemini 3.5 Flash is now the default model powering the Gemini app, AI Mode search, and all other Google AI services, with a more powerful premium iteration, Gemini 3.5 Pro, scheduled to launch next month.
In a rare display of collaboration between two cutthroat competitors, Google also announced that OpenAI will adopt its SynthID invisible watermarking tool for AI-generated images, a joint effort to curb the spread of deepfakes and manipulated AI content across platforms.
For all of Google’s AI advances, the company faces growing pushback from publishers and ongoing legal challenges that threaten its core search business. The expansion of Google’s AI features, which keep users within Google’s own ecosystem rather than directing them to external sites, has raised alarms among news and online publishers who warn the shift will erode their traffic and critical advertising revenue. A 2024 lawsuit filed by media giant Penske Media Corporation – owner of outlets including *Rolling Stone* and *The Hollywood Reporter* – found that 58% of Google searches now end without a user clicking through to any external website.
In Europe, a coalition of major publishing groups has filed a formal complaint with the European Commission, accusing Google of repurposing copyrighted news content to train its AI models and generate AI summaries without compensating publishers for their work. AI Mode remains unavailable in France, the epicenter of a years-long contentious battle between the company and French media outlets over fair compensation for news content.
Legal threats extend beyond the Atlantic as well. A U.S. federal court ruled in 2024 that Google had illegally maintained a monopoly over online search, and the company faces the possibility of being forced to break up core parts of its business. In February, the U.S. Department of Justice appealed a previous ruling that stopped short of ordering Google to divest its popular Chrome browser. Legal experts expect the next hearing in the case to be held no earlier than the end of 2025, with some projections pushing the timeline to 2027.