Tuesday 17 November 2026Updated 18:04 IST
bharat.digital

India's digital economy, public infrastructure and security

Originals · Startup · Startups & Fintech

IIT Kanpur and Bharat1.ai launch AI Kavach, a lab to red team the AI that banks and fintechs deploy

Unveiled at Bengaluru Tech Summit 2026, the lab plans to stress test AI systems used in banking, payments, lending and insurance before they go live, and to train more than 300 AI security specialists a year.

IIT Kanpur and Bharat1.ai launch AI Kavach, a lab to red team the AI that banks and fintechs deploy
Cover: bharat.digital

AI Kavach, an AI security laboratory for India's banking, financial services and insurance sector, is being launched at the 29th Bengaluru Tech Summit, which runs from 17 to 19 November 2026 at the Bangalore International Exhibition Centre. Karnataka IT and BT Minister Priyank Kharge announced the launch at a media conference on the eve of the summit, The Hindu reported, describing it as the country's first AI cybersecurity lab for BFSI. ETV Bharat and CXOToday also list it among the summit's headline launches.

The lab is a joint initiative of IIT Kanpur, which takes part through its WSAIS unit and provides academic leadership, faculty and lab infrastructure, and Bharat1.ai, which says it will supply AI infrastructure, secure testing environments, programme management and deployment platforms. Bharat1.ai is also co-curating the summit's 50,000 sq ft AI and Robotics Pavilion, which features more than 101 startups.

According to the project page, AI Kavach is built on four pillars. The first is controlled red teaming of AI systems used in banking, payments, lending, insurance and wealth management, looking for model failures, prompt injection risks and agentic exploits. The second covers technology to sandbox AI agents, restrict their system access, manage credentials, monitor autonomous behaviour and contain a compromised system quickly. The third is research on assurance, incident attribution, auditability and responsible vulnerability disclosure.

The fourth pillar is people. The lab aims to train more than 300 students and professionals every year through courses, research fellowships and hands-on security exercises, using its own AI compute alongside cloud resources supported through the IndiaAI Mission. The partners say they want to bring in banks, fintechs, AI labs, regulators and security firms, and to publish common methods and evaluation frameworks for securing AI in Indian finance.

The timing matters because AI is moving quickly into fraud detection, customer service, credit decisions and identity checks across Indian finance, where a manipulated model or a rogue agent can move real money. The organisers have not yet named member institutions, a budget or a testing fee, so how widely banks and fintechs use the lab will be the thing to watch.

More originals