Government Taps Sarvam AI, BharatGen to Build Sovereign Cybersecurity AI Models
The move is part of the broader IndiaAI Mission, which aims to reduce dependence on foreign AI models in strategic sectors such as cybersecurity, defence, governance, and public services.
The Government of India has directed homegrown artificial intelligence initiatives Sarvam AI and the IIT Bombay-led BharatGen consortium to develop indigenous AI models for cybersecurity, marking a significant step in the country’s push to strengthen sovereign AI capabilities for critical infrastructure protection.
The latest directive comes amid growing concerns over the use of foreign frontier AI models for sensitive government and national security applications. According to recent reports, the Centre wants domestic AI models capable of supporting cyber threat detection, malware analysis, vulnerability assessment, anomaly detection, and incident response for sectors including defence, banking, telecommunications, power, and government services.
The initiative is being implemented under the ₹10,300 crore IndiaAI Mission, which is building India’s sovereign AI ecosystem through indigenous foundation models, shared compute infrastructure, datasets, startup funding, and AI research.
Earlier this year, the government selected 12 organisations and research consortia including Sarvam AI, BharatGen, Gnani AI, Fractal Analytics, Tech Mahindra Maker’s Lab, and others to develop large and small language models trained on Indian datasets.
Founded in 2023 by Vivek Raghavan and Pratyush Kumar, Bengaluru-based Sarvam AI is developing sovereign large language models tailored for Indian languages, enterprises, and government use cases.
Under the IndiaAI Mission, the startup is building India’s first indigenous foundation model designed for reasoning, voice capabilities, and multilingual applications, with dedicated government-backed computing infrastructure.
Meanwhile, BharatGen, led by IIT Bombay and backed by the Department of Science and Technology, is building multimodal foundation AI models capable of understanding text, speech, and images across more than 22 Indian languages.
The consortium’s models are designed to support applications across governance, healthcare, education, agriculture, and other public-sector use cases while preserving India’s linguistic and cultural diversity.
The government has also significantly expanded AI infrastructure under the IndiaAI Mission by onboarding more than 38,000 GPUs, providing subsidised computing resources to startups, researchers, academic institutions, and public organisations developing indigenous AI technologies.
The latest development underscores India’s growing focus on treating artificial intelligence as strategic national infrastructure. By building sovereign AI models for cybersecurity and other critical domains, the government aims to strengthen digital resilience, improve data sovereignty, and ensure greater control over AI systems powering essential national infrastructure, reducing long-term reliance on overseas AI platforms for sensitive applications.