Artificial Intelligence(AI), is the defining technology of the 21st century. From health, agriculture and education to governance, military and space, the frontier technology is encompassing every domain and is emerging as the core pillar or epicentre of the socio-economic growth trajectory. Artificial Intelligence(AI) is the inconceivable and undeniable reality that has stepped into the human world and is set to dominate the socio-political, economic and cultural landscape.
This tech reality demands a human-centric approach, where the technology is harnessed for human well-being and welfare. Mankind should dictate the terms of the technology for global good and should not be vise versa, where the technology dominates the human world and creates rifts or chaos in the human society in a harmful way. Artificial Intelligence should not be used as a tool of misadventure for nefarious causes such as cyber threats.
Weaponising the frontier technology for geopolitical reasons
Despite this moral underpinning, there is no global consensus for AI governance. The emerging technology is repeatedly used as a geopolitical tool and is weaponised to seek hegemonic ambitions. For example, the US recently curbed access to and imposed export control on Fable 5 and Mythos 5 AI models developed by Anthropic, an US-based AI research and development organisation. As an impact, numerous foreign nationals and entities dependent on the AI model for businesses and other projects were under panic, as the access was denied suddenly.
It was cited that the access was cut due to cyber security and national security concerns. It was to protect the AI models from vulnerabilities. However, many experts asserted that the US has undertaken such a critical decision due to geopolitical reasons, may be to exert hegemonic primacy amid West Asia conflict.
Anthropic had launched next-generation Fable 5 and Mythos 5 recently. Fable 5 is designed 5 general public and has standard API access. It is highly optimized for complex agentic workflows, coding and long-horizon tasks. Meanwhile, Mythos 5 has restricted access and is limited strictly to vetted enterprise partners, government entities and cybersecurity or research professionals.
The sudden denial of access to these AI models by the US, created panic among business and tech entities across the globe, including India, who were dependent on these networks to function. Several IT companies, startups and cyber security researchers in India, have reportedly integrated Fable 5 into their work. However, in July, the US lifted the restrictions imposed on Fable 5 & Mythos 5 and it is again globally accessible.
Yet, this hiccup, reflects the contingencies and vulnerabilities of being dependent on a foreign AI model. It resonates with the importance of developing sovereign AI models, to seal a country’s tech, economic and national security. Similarly, China escalates weaponisation of tech by manipulating exports, bombarding & difussing Chinese AI models in global markets and thereby controlling global data and through other means.
It is an alarm for any nation to develop sovereign AI models and be immune from data theft, cyber threats and other manipulations. Sovereign AI models are critical for economic and larger national security and to insulate from the weaponisation of AI. In this direction, India is marching in the sovereign AI path to secure data localisation, sovereignty and to ensure that the benefits of AI cater to the diverse necessities and unique socio-political landscape of India.
Various entities across India, are developing 12 sovereign AI models, to facilitate AIs credible use for human welfare. The government is facilitating the sovereign AI projects under IndiaAI Mission, with an outlay of Rs. 10, 371 crore. In this, more than Rs. 2,194 crore has been approved to articulate foundational AI models.
Bharat Gen by IIT Bombay
Bharat Gen, is India’s maiden sovereign AI initiative, backed by the government. It is an indigenous multimodal Generative AI initiative that is designed to support 22 official languages of India. Its aim is to build sovereign and language inclusive AI technology to ensure the welfare of the diverse Indian population. Bharat Gen’s core motto is, “Inspired by Bharat, Built for the World”, thus resonating the message of human-centric AI, for global welfare.
Its futuristic approach aims to ensure inclusivity, diversity and authentic Indian representation in the evolving landscape of emerging technologies. BharatGen aims to empower every citizen to engage with the digital world effortlessly. Bharat Gen project says that it aims to bring voice to Bharat through speech recognition and synthesis in Indian languages. It is also building India’s largest open dataset ecosystem to facilitate AI research and innovation.
Sarvam AI
Sarvam is India’s full-stack sovereign AI platform, with speech-to-text, text-to-speech, translation and conversational agents across 22 Indian languages. The Bengaluru-based AI company founded in 2023, develops Large Language Models(LLM) and multimodal AI systems, with a focus on Indian languages and region specific uses, that aims to deeply understand and customise AI to suit the culture and diversity.
The sovereign AI ecosystem developed by Sarvam aims to empower governments, enterprises and other sectors.
Zenteiq
Zentieq, is a Bengaluru-based deep-tech startup that builds native AI infrastructure and scientific foundation models specifically for engineering and industrial R&D. It is backed by the Government’s India AI mission to develop sovereign foundational models for science & research purposes. It aims to harness AI models for complex engineering simulations, thermal analysis, battery design, semiconductor analysis and aerospace systems.
By developing actionable AI, Zentieq aims to accelerate industrial design cycles, reduce computing costs for heavy simulations and bridge the gap between theoretical physics and its application.
Gnani AI
Gnani Ai is a Bengaluru-based deep-tech startup, supported by IndiaAI Mission. It specializes in proprietary speech-to-text, text-to-speech and speech-to-speech models to power multilingual, low-latency enterprise voice agents and automation, to suit the diverse & multilingual Indian landscape. Beyond Indian languages, Gnani Ai can also operate in 40 global languages.
It uses domain-specific Small Language Models(SLMs), rather than bloated LLMs to cut down latency(under 900ms) and lower hallucinations and facilitate real-time interactions.
Socket AI
Socket AI, is an Artificial Intelligence research firm that focuses on developing multilingual foundational models. It aims to build energy-efficient, open-source language models tailored for Indian vernacular languages. The organisation also states that it aims to advance ethical Artificial General Intelligence (AGI) designed for the Global South and real-world deployment.
Fractal Analytics
Fractal Analytics has been selected by the Government of India under the IndiaAI Mission, to develop a sovereign Large Reasoning Model (LRM), focusing specifically on specialized medical reasoning and free healthcare AI services for citizens. Thus, it specialises in medical AI to advance healthcare diagnostics and reasoning capabilities.
Gan AI
Gan AI, under IndiaAI Mission, focuses on developing advanced generative AI and video/text intelligence models, including a 70-billion parameter text-to-speech and personalized video generation framework. It intends to create a real-time, multilingual, emotion-aware voice-to-voice foundational model, tailored for Indian needs.
Intellihealth
Intellihealth is also backed by the IndiaAI Mission, to develop sovereign foundational models for healthcare diagnostic tools. It develops comprehensive AI platforms and open-source technology for government, NGOs and other entities to fulfill the healthcare needs. Most importantly, Intellihealth has developed MANAS-1, India’s first brain language foundation model designed to decode brainwaves and detect neurological and psychiatric disorders.
Avatar AI
Avataar AI under the IndiaAI Mission, is an indigenous project that launched Varya, India’s first distilled AI video-generation model. It creates low-cost, fast text-to-video content tailored for local languages, regional cultures and everyday Indian use cases. It is designed to reflect local communities, diverse languages, festivals and regional Indian contexts. It is specifically built to help MSMEs or small businesses, teachers and other grassroot communities.
Shodh AI
Shod AI, is also an Artificial Intelligence(AI) deep-tech company founded in 2024 that specialises in AI for science. It focuses on foundational physics models, developing high-performance infrastructure for training large neutral networks. For example, under IndiaAI Mission, it launched Project Skanda in collaboration with NVIDIA, which is a mesoscale foundation model for energy storage and material science discovery.
Makers Lab
This is the research and innovation wing of Tech Mahindra and one among the 12 entities chosen by the government under IndiaAI Mission, to develop sovereign AI models. Under the initiative Makers’ Lab has developed an 8-billion-parameter Hindi-first educational Large Language Model(LLM) designed to transform tutoring, STEM learning and multilingual guidance for students.
Genloop
Genloop, under IndiaAI Mission, works with the Government of India, to build LLMs that understands cultural nuances and a diverse language ecosystem.
These 12 sovereign AI projects, backed by the government, are thus powering India’s story in frontier technology. It is scripting an ethical and human-centric AI ecosystem for the country’s and global welfare. India, as the world’s largest democracy and fourth largest economy that is scaling swiftly to be the third largest, is navigating a challenging landscape to carve an ethical and inclusive AI ecosystem amid attempts of weaponization.
