Google doubles down on India’s AI ambitions with enterprise AI, safety tools and education push at I/O Connect 2026

Google used its I/O Connect India 2026 event to outline the next phase of its India AI strategy, shifting the conversation beyond models to enterprise deployment, developer tools, AI safety and education. The company announced new products and partnerships aimed at helping businesses, governments and developers build AI applications while addressing trust, security and data localization requirements.

ET was on the ground at Google I/O Connect India 2026, where Google executives showcased several new AI capabilities alongside live demonstrations from Google DeepMind. ET also spoke to Preeti Lobana, VP and Country Manager, Google India, on the pace of AI adoption in the country. The conversation will be published soon.

Opening the event, Karnataka Deputy Chief Minister D.K. Shivakumar highlighted the state’s long-standing partnership with Google and invited the company to make Karnataka a testbed for future AI innovation. Google, meanwhile, positioned India as one of its fastest-moving AI markets, with Preeti Lobana saying the company is focused on building the infrastructure and guardrails needed as AI moves into the “agentic” era, where systems perform tasks rather than simply answering questions.


A major enterprise announcement was the expansion of Gemini on Google Distributed Cloud, allowing regulated industries and public sector organisations to deploy Google’s frontier AI models within their own infrastructure and Indian data centres. The move is designed to address growing concerns around compliance, data residency and security as enterprises adopt generative AI.

Google also introduced a series of security initiatives centred around agentic AI. These include extending Sec-Gemini V3, its AI-powered cybersecurity agent, to trusted enterprise and government testers in India, while open-sourcing CAPSEM, a secure runtime environment that isolates AI agents inside virtual machines to reduce the impact of compromised agents or malicious prompts. The company is also backing new open standards for secure AI payments, device protection and agent-to-agent communication.

Education remained another major theme. Google DeepMind announced AI Research Foundations, a free 56-hour curriculum designed to teach developers how to build and fine-tune large language models. The programme will be offered through Google’s learning platforms in partnership with NASSCOM and IISc Bengaluru. The company also introduced ATL Saathi, a Gemini-powered assistant for teachers working with Atal Tinkering Labs, beginning with 100 schools before expanding further.

On the healthcare front, Google said researchers at AIIMS Delhi are using its MedGemma open models to build India-specific AI models focused on leprosy and sexual and reproductive health, with the resulting models expected to be made available to the broader developer ecosystem. Gemini Live is also expanding support to more than 25 Indian languages and dialects, including Sanskrit, Bhojpuri and Maithili.

Google also highlighted the economic impact of its ecosystem in India, citing third-party research that estimates Google Play and Android generated ₹5.3 lakh crore in revenue for Indian app publishers and the wider digital economy in 2025, a 28% increase over the previous year. The company also announced plans to bring its Google Play Academy curriculum to 10,000 developers in partnership with multiple state governments.

Beyond the keynote, the event featured live demonstrations from Google DeepMind, including a “Sketch to Reality” workflow where a hand-drawn interface was converted into a working UI prototype in seconds using multiple Google AI models.ET also spoke with Dr. Manish Gupta, Senior Director, Google DeepMind, and Seshu Ajjarapu, Senior Director, Applied AI, about India’s AI ecosystem, enterprise adoption and the next wave of agentic AI. These conversations will be published soon.