How India's AI Policy and IndiaAI Mission Work
India's AI governance architecture has two complementary tracks: one promotes AI adoption and builds domestic AI capabilities through the IndiaAI Mission; the other governs AI risks through guidelines, the DPDPA framework, and sectoral regulatory requirements.
The IndiaAI Mission — approved by Cabinet on March 7, 2024, with a budget of ₹10,372 crore ($1.25 billion) — is the primary investment vehicle, structured around seven pillars: AI compute infrastructure (targeting 10,000+ GPUs through public-private partnership, with 34,000+ GPUs deployed by May 2025); IndiaAI Innovation Centre for indigenous AI model development; the IndiaAI Dataset Platform for open data access; FutureSkills for AI skills development; Safe and Trusted AI for governance and ethics; AI in Startups for risk capital; and AI for India for socially impactful applications.
 |
| Representational Image: How India's AI Policy and IndiaAI Mission Work |
The government confirmed in Parliament (December 2024) it
was taking a "voluntary, self-regulatory approach" to AI, with
industry body NASSCOM tasked with developing non-binding guidelines. India does
not plan to introduce standalone AI legislation, instead adapting the DPDPA, IT
Act, and existing IP laws to address AI-specific concerns. The guidelines apply
seven "sutras" (principles): Trust, People First, Innovation over
Restraint, Fairness and Equity, Accountability, Transparency, and Privacy.
What You Need to Know
- IndiaAI
Mission: Cabinet-approved March 7, 2024; ₹10,372 crore ($1.25 billion)
budget; seven pillars; 34,000+ GPUs deployed by May 2025; AI Safety
Institute (AISI) proposed as central safety testing and research body; AI
Governance Group (AIGG) as permanent inter-agency coordination body;
Technology and Policy Expert Committee (TPEC) for technical advice.
- AI
Governance Guidelines (November 5, 2025): voluntary, non-binding; seven
principles ("sutras"); "lightweight" and adaptive
regulatory approach; proposes AI Governance Group (AIGG), AI Safety
Institute (AISI), graded liability approach; advocates
"whole-of-government" coordination with sectoral regulators
(RBI, SEBI, TRAI) incorporating AI governance into domain regulations.
- India's
AI ranking: Stanford AI Index Report 2025 ranked India second in countries
with highest AI skill penetration (2015–2024); top 10 in private AI
investment globally (2013–2024); India chaired the Global Partnership for
AI (GPAI) in 2024.
- No
standalone AI law: MeitY confirmed December 2024 that India will not
introduce standalone AI legislation; existing DPDPA, IT Act, and IP laws
will apply to AI; Digital India Act — proposed successor to IT Act 2000 —
remains unintroduced as of May 2026.
- MeitY
advisory (March 15, 2024): required intermediaries to ensure AI models
deployed in India carry labels indicating potential inaccuracies;
initially required government approval for "under-testing" AI
models; government approval requirement removed after industry opposition;
labelling requirement remained; represents India's first specific AI
compliance obligation for platforms.
How It Works in Practice
1. The AI compute gap and IndiaAI's role: India's AI
ambitions have historically been limited by insufficient domestic compute
infrastructure; most Indian AI development used cloud compute from US
hyperscalers (AWS, Azure, GCP). The IndiaAI Mission's GPU deployment target —
10,000+ GPUs initially, 34,000+ deployed by May 2025 — aims to create a
domestic compute pool accessible to Indian researchers, startups, and
institutions at subsidised rates, reducing dependence on foreign infrastructure
and enabling computation of sensitive data within India.
2. The AI Safety Institute's role: The proposed AI
Safety Institute (AISI) — not yet fully operational as of May 2026 — is
envisioned as a hub-and-spoke institution that conducts safety testing of AI
systems, develops India-specific risk frameworks, coordinates with sectoral
regulators, and represents India in international AI safety forums. The AISI
model is influenced by the UK's AI Safety Institute (AISI), although India's
version is proposed as an advisory and research body rather than a mandatory
pre-deployment testing authority.
3. Sectoral AI regulation emerging: Sectoral
regulators are developing their own AI frameworks without waiting for central
AI legislation. The RBI's FREE-AI (Framework for Responsible and Ethical
Enablement of AI) Committee report (August 2025) proposed specific AI
governance requirements for the financial sector; SEBI has issued AI-related
guidance for securities market participants; TRAI is examining AI in telecom.
This sectoral fragmentation creates a patchwork regulatory environment that the
AIGG is supposed to coordinate.
4. AI for India's public services: The IndiaAI
Mission's "AI for India" pillar targets socially impactful
applications: AI in healthcare (disease diagnosis), agriculture (crop advisory,
AgriStack integration), education (personalised learning), governance (public
service delivery optimisation), and language (BHASHINI expansion). Government
AI deployments — including DoordarshanKisan's AI anchors, judicial AI tools for
case management, and DBT targeting algorithms — represent the real-world AI
governance challenge of ensuring government-deployed AI systems meet
accountability standards.
5. The Grok incident as a governance stress test: The
Grok NCII incident (December 2024) demonstrated that India's current regulatory
framework — routing AI-generated harm through platform intermediary liability —
cannot adequately address harms originating from AI systems that are not
primarily social media intermediaries. TechPolicy.Press analysis identified: no
pre-deployment testing requirements for AI systems in India; no mandatory
safety guardrails; no independent oversight mechanism for high-risk generative
AI; regulation only triggers after harm occurs through intermediary liability,
not before.
What People Often Misunderstand
- India's
innovation-first approach is deliberate, not a regulatory gap: India
has consciously chosen not to regulate AI technology itself, only harmful
applications; this is a policy choice reflecting India's position as an AI
user rather than a frontier AI developer, wanting to adopt AI tools
without imposing restrictions on adoption speed.
- NASSCOM's
voluntary guidelines are not binding: Industry self-regulation through
NASSCOM guidelines is the current accountability mechanism for AI
governance in India; voluntary guidelines without enforcement are less
effective than statutory requirements but impose lower compliance costs;
the balance is a policy choice.
- India's
AI governance competes for attention against AI investment promotion:
The same ministry (MeitY) that governs AI risk also promotes AI adoption
through IndiaAI Mission; this institutional combination creates incentives
to favour innovation promotion over risk governance.
- The
EU AI Act model is not India's reference point: India's
"innovation over restraint" philosophy is explicitly a contrast
to the EU AI Act's risk-based mandatory requirements; India's approach
shares more with the US sectoral framework than the EU comprehensive
model; comparing India unfavourably to the EU AI Act misunderstands
India's deliberate policy stance.
- Deepfakes
and election AI present genuine unresolved regulatory challenges: The
Grok incident and election deepfake proliferation demonstrate that India's
current framework leaves significant harmful AI applications in regulatory
grey zones; these will require either new legislation or specific sectoral
rules as harms accumulate.
What Changes Over Time
The AISI's first technical standards for safety testing — expected in late 2026 — will define India's first AI safety benchmarks; their specific requirements will shape how AI developers approach the Indian market.
The Digital India Act (DIA) — if eventually introduced and passed — would
create a more comprehensive regulatory framework for digital services including
AI; its indefinite postponement reflects political difficulty in legislating
technology-specific norms at scale.
Sources and Further Reading
- PSA
— AI Mission and Initiatives: https://www.psa.gov.in/ai-mission-initiatives
- IAPP
— India AI Governance Guidelines: https://iapp.org/news/a/notes-from-the-asia-pacific-region-india-releases-dpdpa-rules-ai-governance-guidelines
- IAPP
— Global AI Governance India: https://iapp.org/resources/article/global-ai-governance-india
- MediaNama
— India AI law stalled: https://www.medianama.com/2025/12/223-india-ai-law-digital-india-act-stalled/
- Saikrishna
& Associates — India AI Governance Guidelines: https://www.saikrishnaassociates.com/decoding-the-india-ai-governance-guidelines/