AI Governance in India: The National Strategy and the Regulatory Gap
May 27, 20267 min read
The question reads: "Consider the following statements about India's approach to Artificial Intelligence governance."
You know about ChatGPT. You know about DeepSeek. You know AI is changing everything. But the exam question is not about the technology. It is about the policy framework.
"A 2018 strategy document. No legislation enacted. Multiple sectoral regulators claiming jurisdiction." This describes India's AI governance architecture - or rather, the absence of one. The EU passed the AI Act in 2024, the world's first comprehensive AI law. The US issued an Executive Order on AI in 2023. China has implemented algorithmic recommendation regulations, deep synthesis rules, and generative AI measures. India has a National AI Strategy from 2018 that recommends no regulation at all.
[TOPIC CLASSIFICATION]
Topic type: Science & Technology Policy (governance of emerging technology)
PYQ frequency: Low-Medium (1-2 questions in last 3 years, rapidly increasing)
Trap: Confusing mission objectives, launch vehicles, or technological capabilities. Examiners test precise technical terminology.
Most confused: The distinction between civilian and strategic (dual-use) technology, and between indigenous development vs licensed production vs import.
Key anchor: The institutional architecture (ISRO, DRDO, DAE, CSIR, ICMR, DBT, MeitY) and policy framework (Space Policy, Nuclear Doctrine, Biotechnology Guidelines, National AI Strategy, Drone Rules).
Current affairs hook: Major missions (Chandrayaan-3, Aditya-L1, Gaganyaan), new policy initiatives (IN-SPACe, iDEX, Semiconductor Mission, Quantum Mission, AI Mission), defence tests (Agni, S-400, ARIHANT), health tech (mRNA, vaccines).
Mains hinge: Frame answers around the core tension - self-reliance vs international cooperation, strategic autonomy vs technology access, security vs transparency, innovation vs regulation.
Core Concept
Artificial Intelligence governance refers to the policies, laws, regulations, and institutional mechanisms that guide the development and deployment of AI systems. Unlike traditional technology regulation (which regulates hardware or software products), AI governance regulates outcomes - what the AI system does, not what it is.
India's AI governance journey:
2018: NITI Aayog releases "National Strategy for AI" (#AIforAll). The strategy is framework-oriented, not regulatory. It recommends leveraging AI for social good and creating a two-tier institutional structure.
2020: NITI Aayog launches the Responsible AI framework - a series of discussion papers on transparency, accountability, bias, and ethics in AI. Non-binding.
2023: The proposed Digital India Act (replacing IT Act 2000) includes provisions for AI regulation. Not yet introduced in Parliament.
2024: Ministry of Electronics and IT (MeitY) issues advisory requiring government approval for AI model launches. Withdrawn after backlash. Replaced with a "label it" advisory - AI-generated content must be labelled.
2025: National AI Mission approved with 10,372 crore outlay. Focuses on compute infrastructure (10,000+ GPUs), AI datasets, skill development, and sectoral AI applications. Does not include regulatory components.
India's institutional landscape for AI is fragmented:
MeitY: Overarching policy and digital regulation
NITI Aayog: Strategy and responsible AI framework
DST: Fundamental AI research funding
Sectoral regulators: RBI (AI in fintech), IRDAI (insurance AI), TRAI (AI in telecom), SEBI (AI in markets) - each developing their own approaches
Ministry of Defence: AI in defence applications (Defence AI Council, 2022)
Ministry of Home Affairs: AI for internal security (facial recognition, predictive policing)
The global landscape:
EU AI Act (2024): Risk-based framework - unacceptable risk (banned), high risk (mandatory compliance), limited risk (transparency), minimal risk (no obligations). Fines up to 7% of global turnover.
US Executive Order (2023): Requires safety testing for AI models, watermarking of AI content, privacy protection, and equity assessment. Applies to federal agencies and companies developing large-scale AI models.
China: Algorithmic recommendation regulations (2022), deep synthesis rules (2023), generative AI measures (2023). Requires AI models to reflect "core socialist values."
India: No comprehensive regulation. Sectoral guidance only. The "light-touch" approach favours innovation but creates regulatory uncertainty.
The key debate in India is whether to regulate the model (the AI system itself) or the application (what the AI is used for). The EU regulates models above a compute threshold (training compute > 10^25 FLOPs). India's approach has been to regulate applications (e.g., AI in lending regulated by RBI, AI in healthcare by CDSCO). The absence of a horizontal framework means gaps exist where no sectoral regulator claims jurisdiction.
Key Facts
India's AI market size: $7.8 billion (2024), projected $17 billion by 2027 (25% CAGR)
National AI Mission outlay: ₹10,372 crore (2025-2030)
AI compute infra target: 10,000+ GPUs under National AI Mission
AIRAWAT supercomputer: ranked 75th globally (2024), deployed at C-DAC Pune
INDIAai portal: national AI knowledge platform (MeitY + NeGD)
India's AI start-ups: 4,200+ (third largest globally)
Global AI regulation: EU AI Act (enforced from 2025), US Executive Order (2023), China's Generative AI Measures (2023)
Proposed Digital India Act: expected to replace IT Act 2000, includes AI regulation - still pending
India's National Strategy for AI (2018): five focus sectors - healthcare, agriculture, education, smart cities, mobility/infrastructure
PM-SHRI (PM's Science, Technology, and Innovation Advisory Council): apex body for AI policy coordination
India's AI workforce: 1.2 million (2025), projected 2 million by 2028
Previous Year Questions
Year
Stage
What was tested
2024
Prelims
AI vs Machine Learning vs Deep Learning definitions
2023
Prelims
Applications of AI in Indian agriculture
2024
Mains GS-3
"The regulation of AI poses a dilemma between fostering innovation and ensuring safety." Discuss with reference to India.
2022
Mains GS-3
"AI is poised to transform the Indian economy." Examine the potential and challenges.
2021
Prelims
INDIAai portal and National AI Strategy
2020
Prelims
AIRAWAT supercomputer location
Statement Elimination Guide
"India has passed comprehensive AI regulation through the Digital India Act." False. The Digital India Act has been proposed but not introduced in Parliament. India has no comprehensive AI regulation.
"The DPDP Act 2023 regulates AI systems that process personal data." False. The DPDP Act regulates personal data processing. An AI system trained on anonymised data is not covered. The DPDP Act is a data protection law, not an AI governance law.
"The EU AI Act uses a risk-based approach to classify AI systems into four categories." Correct. Unacceptable risk (banned), high risk (mandatory compliance), limited risk (transparency obligations), minimal risk (no obligations).
"India's National AI Strategy recommends mandatory regulation for all AI applications." False. The 2018 strategy recommends self-regulation for medium-risk AI and only mandatory compliance for high-risk AI in five identified sectors.
"The National AI Mission includes both compute infrastructure and regulatory components." False. The AI Mission (2025, ₹10,372 crore) focuses on compute, datasets, and skills. It does not include regulatory components - regulation remains with MeitY.
Current Affairs Hook
The 2024 advisory that required AI models to seek government approval before public release was criticised as creeping censorship and impractical for open-source models. The government backtracked within three weeks. The replacement advisory only requires "labelling of AI-generated content" and "explicit permission before using copyrighted training data." The episode revealed the government's uncertainty about AI regulation.
The 2025 OECD AI Policy Observatory report noted that India is one of the few major economies without a dedicated AI regulatory body. The report recommended establishing an independent AI Safety Authority within MeitY with powers to conduct audits, certify compliance, and impose penalties.
The issue of AI-generated deepfakes in elections became prominent during the 2024 general election. MeitY issued an advisory under Section 66D of the IT Act (punishment for cheating by impersonation using computer resources) - treating deepfakes as an existing cybercrime rather than a new AI-specific category. The Election Commission did not issue AI-specific campaign guidelines for the 2024 election.
India's chairmanship of the Global Partnership on Artificial Intelligence (GPAI) in 2024 focused on trustworthy AI development, data governance, and AI for sustainable development. The GPAI summit in New Delhi (2024) adopted a ministerial declaration on "Responsible AI for All" but produced no binding commitments.
Interlinkages
Ethics: Algorithmic bias - AI systems trained on Indian data can reflect and amplify existing social biases (caste, gender, religion). The responsible AI framework (NITI Aayog) discusses fairness and non-discrimination but provides no enforcement mechanism.
Economy: AI adoption in Indian banking (credit scoring, fraud detection), insurance (claims processing), and agriculture (crop advisory, price prediction). The Economic Survey 2023-24 included a chapter on AI's impact on the Indian labour market, estimating 30-40% of routine white-collar jobs could be affected.
Internal Security: Facial recognition systems deployed by Delhi Police (2009 onwards) and other state police forces. No central law regulates police use of AI. The Supreme Court Puttaswamy right-to-privacy judgement applies but has not been interpreted for AI surveillance.
Governance: AI in public service delivery - UMANG app, Aadhaar-based authentication, AI chatbots for scheme information. The government's AI-for-All approach positions AI as a governance tool rather than a regulatory problem. This framing shapes the policy response.
International Relations: India's AI data localisation requirements (under PDPA 2019 draft, softened in DPDP 2023) affect global AI companies training models on Indian data. The US-India Initiative on Critical and Emerging Technology (iCET, 2023) includes AI cooperation - joint research, standard-setting, and talent exchange.
Common Mistakes
Assuming India has AI regulation because it has a data protection law. The DPDP Act 2023 and AI governance are separate. India has one, not the other.
Confusing the National AI Strategy (2018, NITI Aayog) with a regulatory framework. The strategy is a policy document, not a law. It has no enforcement provisions.
Treating government advisories on AI as binding law. The 2024 advisory requiring government approval for AI models was withdrawn. Even when active, advisories are not laws - they are executive instructions.
Overlooking the sectoral fragmentation. AI in banking is regulated by RBI, AI in insurance by IRDAI, AI in capital markets by SEBI. There is no single AI regulator. This creates regulatory gaps and overlaps.
Thinking the EU AI Act applies globally. The EU AI Act applies to AI systems deployed in the EU market. Indian companies exporting AI services to the EU must comply. Indian companies operating only in India are not covered - unless India passes equivalent legislation.
Revision Snapshot
India has no comprehensive AI regulation. The National AI Strategy (2018, NITI Aayog) recommends a risk-based approach but has no legal force. The proposed Digital India Act (expected to include AI provisions) has not been introduced. Sectoral regulators (RBI, SEBI, IRDAI) govern AI in their domains, creating a fragmented landscape. The National AI Mission (2025, ₹10,372 crore) focuses on compute infrastructure and skills, not regulation. Globally, the EU AI Act (2024) is the first comprehensive AI law; the US relies on executive action; China has targeted AI regulations. India's "light-touch" approach favours innovation but creates uncertainty. The key UPSC takeaway: India has an AI strategy (2018), an AI mission (2025), and fragmented sectoral regulation - but no AI law.
Source Notes
Department annual reports (Space, Atomic Energy, Defence R&D, Biotechnology, S&T)
ISRO, DRDO, DAE, CSIR, ICMR, DBT, MeitY official publications