India’s AI Gamble: A Billion Users and No Rulebook
India has quietly become the world’s largest AI market without passing a single AI law. For some, that is visionary. For others, it is a disaster waiting to happen. Here is the full story.
Series: Part One of Three | India’s AI Revolution
Category: Technology | Policy | India
Estimated Read Time: 4 minutes
Word Count: ~950 words
On a Tuesday morning in a village in Tamil Nadu, a farmer picks up his phone and speaks his question aloud. Within seconds, an AI chatbot tells him exactly why his crop insurance payment has been delayed and what he needs to do. No office visit. No form. No middleman. Just an answer, in Tamil, in his own voice.
This is what artificial intelligence looks like in India. Not robot lawyers or self-driving cars. Not chatbots writing poetry for bored office workers in San Francisco. It looks like a farmer in Tamil Nadu getting an answer he could never get before.
India is in the middle of the most ambitious, least understood and arguably most consequential AI experiment on the planet. And it is doing this without the sprawling, heavily penalised regulatory framework that Europe spent years constructing.
The Decision Not to Write a Law
India has looked at the European Union’s AI Act, with its hundreds of pages of regulation and strict risk classifications, and made a deliberate decision to go another way.
India’s IT Secretary S. Krishnan explained at the launch of the country’s AI Governance Guidelines in November 2025 that India had “consciously chosen not to lead with regulation but to encourage innovation while studying global approaches.”
The government calls this a “light touch” approach. In practice, it means India governs AI through seven core principles rather than enforceable law. Those principles, called the Seven Sutras, are: Trust is the Foundation. People First. Innovation over Restraint. Fairness and Equity. Accountability. Understandable by Design. Safety, Resilience and Sustainability.
They are not laws. Companies are expected to follow them. The implicit message from New Delhi is that binding law could follow if they do not. The Guidelines establish a whole institutional architecture to support this approach. An AI Governance Group, chaired by the Principal Scientific Adviser, coordinates across government.
A Technology and Policy Expert Committee provides technical advice. A newly created AI Safety Institute monitors emerging risks. And sectoral regulators like the Reserve Bank of India and the Securities and Exchange Board of India retain full enforcement powers within their domains. This is not a vacuum. It is a distributed governance model designed to be adaptive rather than monolithic. But the Guidelines are explicitly described as a strategic window for responsible self regulation. Mandatory binding rules are not ruled out. They are deferred.
The Three Hour Rule That Made History
On 10 February 2026, India did something no country had done before. It passed the world’s fastest mandatory content removal timeline for AI generated harmful material.
Platforms now have three hours to remove deepfakes and synthetically generated harmful content once they receive a lawful government order. For nonconsensual intimate imagery, the deadline is two hours. The previous window was 36 hours.
To understand why India moved this fast, consider what was happening. Deepfake related cybercrime in the country increased by 550 percent since 2019. A morphed video of a Bollywood actress sparked national outrage in 2023. AI cloned voices of politicians spread false voting information during the 2024 general election. A 19 year old student in Faridabad died by suicide after being blackmailed with deepfake images.
The same February 2026 amendments did more than just accelerate timelines. They defined “synthetically generated information” in Indian law for the first time. They mandated platforms to prominently label AI generated content and embed permanent metadata for traceability. A “Generated by AI” tag is now mandatory, not optional. Intermediaries must also issue quarterly compliance notifications to users and acknowledge grievances within seven days, with final resolution required within thirty six hours. The regulatory direction is more comprehensive than a single headline number suggests.
There is, however, an honest gap. The deepfake law contains no explicit protection for satire or parody. A realistic looking satirical video of a politician could legally be treated the same way as malicious disinformation. No court has drawn that line yet.
The Money Behind the Mission
India’s government is not only writing principles. It is spending money. In March 2024, the government approved the IndiaAI Mission, allocating approximately 1.25 billion US dollars over five years to build India’s sovereign AI infrastructure.
As of August 2025, over 34,000 high powered AI chips called GPUs have been made available to Indian startups and researchers at subsidised rates, with a target of reaching 100,000. India’s national AI resource platform hosts datasets and AI models built specifically for Indian conditions.
Four Indian companies are developing homegrown AI models with government support. One of them, Sarvam AI, launched both a 35 billion and a 105 billion parameter multilingual model at the India AI Impact Summit in February 2026. The government gave Sarvam access to 4,096 GPUs and subsidised training costs estimated at nearly 100 crore rupees. It is a significant national achievement, though the model has not been released on open platforms, meaning independent experts cannot yet verify its performance claims.
How AI Is Actually Changing Lives
The numbers matter. But so do the individuals behind them.
Kisan-eMitra, the government’s AI chatbot for farmers, handles queries every single day across 11 regional languages. It helps farmers check their income support payments, verify crop insurance status and understand their loan eligibility, entirely by voice.
DIKSHA, the national digital education platform, uses AI to read textbooks aloud for students with visual impairments and delivers lessons across 36 Indian languages. It has reached 275 million users, with a significant majority coming from rural areas.
In healthcare, AI diagnostic tools are giving small clinics in remote villages access to the kind of specialist analysis previously available only in large city hospitals. The government has established Centres of Excellence in healthcare, agriculture, sustainable cities and education specifically to scale these applications.
The Honest Concern
Not everyone is convinced this experiment is going well.
Critics point out that India’s Seven Sutras carry no legal penalty. A company that ignores every single one of them faces no fine, no prosecution and no mandatory audit.
The World Economic Forum has flagged India as among the countries facing severe risk from misinformation and disinformation. Voluntary principles have not yet answered that challenge.
A private member’s bill called the AI Ethics and Accountability Bill was introduced in India’s Parliament in December 2025, proposing mandatory ethics reviews and fines of up to five crore rupees for violations. It has not yet passed. But its very existence signals that even inside Parliament, there is growing unease about whether self regulation is enough.
The Verdict
India is attempting something genuinely new. Not Europe’s precautionary model, which produced comprehensive law but slow adoption. Not America’s hands off model, which produced rapid innovation but patchy protection. Something in between, designed specifically for a country of 1.4 billion people, 22 languages and a farming population that still has more use for a crop insurance chatbot than for a large language model that writes haiku.
According to the Stanford Artificial Intelligence Index Report 2025, India now ranks second globally in AI skill penetration. The country’s AI market is projected to grow significantly by 2027.
The Tamil Nadu farmer talking to his phone. The student in Rajasthan hearing her textbook read aloud. The informal worker finding a job through an AI powered platform. These are not pilot projects from a government whitepaper. They are running at scale, right now, for hundreds of millions of people.
Whether India can sustain that while also protecting its citizens from the deepfakes, disinformation and data exploitation that travel alongside it, that is the question that defines the next decade.
The experiment is running. The results are coming in…
This article is based on primary sources including the MeitY AI Governance Guidelines (November 2025), the IT Amendment Rules 2026, the Digital Personal Data Protection Rules 2025, the IndiaAI Mission reports and the Stanford Artificial Intelligence Index Report 2025. All figures reflect verified data as of April 2026.
This is Part One of a three part series on India’s AI revolution.
Part Two: India AI Landmine: What Every UK and US Founder Must Know Before Entering the World’s Largest AI Market
Part Three: India’s AI Story: The Version Journalists Are Missing
This article is based on primary sources including the MeitY AI Governance Guidelines (November 2025), the IT Amendment Rules 2026, the Digital Personal Data Protection Rules 2025, the IndiaAI Mission reports and the Stanford Artificial Intelligence Index Report 2025. All figures reflect verified data as of April 2026.
Submit your review | |