Google localises Gemini AI for regulated sectors in India, expands enterprise push with offline deployment

Google announced at its I/O Connect India 2026 event in Bengaluru that Indian enterprises and public sector organizations will be able to run Gemini AI models within Indian data centers through Google Distributed Cloud. The move marks a significant step in Google's push into regulated sectors requiring data localization and tighter security controls.

The company said enterprises, including those in regulated industries, can deploy Gemini entirely within their own infrastructure, with supporting services operating without a connection to the public internet. Gemini 3.5 Flash will also be available with in-country machine learning processing commitments for Indian enterprises and startups through the Gemini Enterprise platform.

"India's builders are already deploying AI faster than almost anywhere else," said Preeti Lobana, Country Manager for Google India. "As we drive the shift into the agentic era, where AI moves from answering queries to securely executing tasks, our focus is on providing the underlying infrastructure and guardrails the ecosystem needs to scale safely."

Alongside the enterprise announcements, Google introduced new security tools for agentic AI. The company said Sec-Gemini V3, its cybersecurity agent, will be made available to selected government and enterprise testers including Flipkart. Google also announced CAPSEM, an open-source secure runtime environment that isolates AI agents inside virtual machines to limit the impact of security breaches or malicious prompts.

Google also announced open standards for secure AI-based transactions and inter-agent collaboration. Device Bound Session Credentials (DBSC) protects against stolen session cookies, while Agents-to-Payments enables authorized low-value AI-led financial transactions alongside the Agent2Agent protocol. The company said it has partnered with IIT Delhi and IIT Madras on research into agentic AI safety and threat detection.

In education, Google DeepMind launched AI Research Foundations, a 56-hour program on building and fine-tuning large language models, in partnership with NASSCOM and the Indian Institute of Science. It also introduced ATL Saathi, a Gemini-powered assistant for teachers in Atal Tinkering Labs, initially rolling out to 100 schools with plans to expand to 10,000.

In healthcare, researchers at AIIMS Delhi are using Google's MedGemma models to develop AI applications for leprosy and sexual and reproductive health, with the resulting clinical models to be made available to the Indian developer ecosystem. Gemini Live has also been expanded to support more than 25 Indian languages and dialects.

During the event, Google noted that its Play and Android ecosystem generated an estimated Rs 5.3 lakh crore ($64 billion) in revenue for app publishers and the broader Indian economy in 2025, a 28% increase from the previous year.