What Happened
- Analysts and policy experts have observed that the dominant norm in global AI governance frameworks is currently performance and scale — the ability to build and deploy larger, more capable AI systems faster — rather than safety, accountability, or equity.
- India's 2025 AI Governance Guidelines, released by the Ministry of Electronics and Information Technology (MeitY), took an innovation-first approach structured around "Seven Sutras" and six pillars of governance, positioning India as a potential rule-setter for the Global South.
- Critics argue that while India's framework is thoughtful domestically, India needs to be more assertive in multilateral forums — pushing back against frameworks dominated by the US and China that optimise for capability and competitive advantage rather than protecting vulnerable populations or ensuring equitable access.
- India's February 2026 AI Impact Summit generated debate about whether India's AI governance posture reflects its development priorities or is inadvertently aligning with the performance-first paradigm of advanced AI nations.
- The article argues India must leverage its Digital Public Infrastructure (DPI) experience — Aadhaar, UPI, DigiLocker — and its representation of 1.4 billion people to demand that global AI standards incorporate safety, equity, and the interests of the Global South.
Static Topic Bridges
Global AI Governance Landscape
AI governance refers to the rules, norms, standards, and institutions that shape how AI is developed, deployed, and regulated. The governance landscape is fragmented, with major geopolitical blocs pursuing different models.
- EU AI Act (2024): The world's first comprehensive binding AI law; risk-based approach — bans high-risk applications (biometric surveillance, social scoring), mandates transparency and human oversight for general-purpose AI; sets a global benchmark
- US approach: Executive Order on AI (2023, Biden era); currently under revision under the Trump administration, which rolled back some AI safety requirements; prioritises innovation and national competitiveness
- China: Has sectoral AI regulations (generative AI regulations 2023, algorithmic recommendation rules); emphasises state control, content moderation, and technological self-sufficiency
- G7 Hiroshima AI Process: Produced voluntary international guiding principles and code of conduct for advanced AI developers (2023); non-binding
- UN AI Resolution (2024): First UN General Assembly resolution on AI, co-sponsored by 123 countries including India; calls for safe, secure, and trustworthy AI that respects human rights and supports sustainable development
- Paris AI Action Summit (February 2025): French-hosted summit produced a "Declaration on Inclusive and Sustainable AI"; India signed
Connection to this news: The article's argument is that these major frameworks — particularly the US and China models — are converging on capability competition ("AI arms race") rather than safety and equity, leaving Global South interests underrepresented.
India's AI Governance Guidelines (2025)
India's AI Governance Guidelines, released by MeitY in late 2025, are structured around a seven-principle framework adapted to India's development context and DPI experience.
- Seven Sutras: Trust as Foundation; People First; Fairness & Equity; Accountability; Understandable by Design; Safety / Resilience / Sustainability; Innovation over Restraint
- Six Pillars: Infrastructure; Capacity Building; Policy & Regulation; Risk Mitigation; Accountability; Institutions
- Risk classification: focuses on national security issues and harms to vulnerable groups (deepfakes targeting women, child safety, language and caste bias in AI outputs)
- National AI Incident Database: proposed mechanism to track, learn from, and disclose AI failures — a transparency tool modelled on aviation incident reporting
- Regulatory sandbox: allows AI developers to test products in controlled environments before full deployment
- India's framework is largely voluntary/advisory rather than binding at this stage — contrasting with the EU's hard law approach
Connection to this news: India's domestic framework contains strong equity and safety language, but critics argue this has not been matched by aggressive advocacy in multilateral settings where global AI norms are actually being set.
India's Digital Public Infrastructure (DPI) as a Model
India's DPI — a stack of open, interoperable digital systems built on public infrastructure — is seen globally as a model for technology-enabled governance that can be exported to other developing nations. It provides India with a distinctive voice in global tech governance conversations.
- Aadhaar: Biometric digital identity for 1.4 billion Indians; world's largest biometric identification system; managed by UIDAI
- UPI (Unified Payments Interface): Real-time payment system; processed ~100–130 billion transactions in FY2024; interoperable; exported to Singapore (PayNow), UAE, France, others
- DigiLocker: Digital document storage linked to Aadhaar; 250+ million users
- ONDC (Open Network for Digital Commerce): Open protocol for e-commerce interoperability — anti-monopoly alternative to closed platforms like Amazon/Flipkart
- India Stack: Collective term for Aadhaar + eKYC + UPI + DigiLocker as a layered DPI architecture
- India has proposed the DPI model at the G20 and at the UN as a template for Global South countries to leapfrog legacy infrastructure
Connection to this news: India's argument in global AI governance could be grounded in DPI experience: open, interoperable, publicly governed technology infrastructure can serve equity goals that closed, proprietary AI systems cannot.
AI and the Global South — Equity Concerns
The concentration of AI development capacity in the US, EU, and China raises equity concerns: that AI systems will be designed for majority-language, high-income users and may embed biases that harm lower-income, multilingual, or marginalised populations.
- Over 7,000 languages exist globally; most frontier AI models are trained predominantly on English-language data
- India has 22 scheduled languages; AI systems that perform poorly in Hindi, Tamil, Bengali, or other Indian languages disadvantage hundreds of millions of users
- AI-driven automation may disproportionately affect jobs in lower-income countries that currently rely on labour-intensive manufacturing and services
- Compute inequality: training frontier AI models requires billions of dollars in GPU compute, available only to large corporations in wealthy nations; cloud access costs further restrict Global South AI development
- India's Bhashini project: National Language Translation Mission to build AI tools for all 22 scheduled languages — an equity-oriented DPI initiative
Connection to this news: The article argues that India should push for "compute equity" provisions in global AI frameworks — ensuring developing nations have access to AI infrastructure rather than being mere consumers of AI systems built by others.
Key Facts & Data
- EU AI Act: came into force August 2024; full application from August 2026; first binding AI legislation globally
- UN UNGA AI Resolution: passed March 2024; co-sponsored by 123 countries including India; first UNGA resolution on AI
- India's AI Governance Guidelines: released by MeitY, late 2025; structured around Seven Sutras and six pillars
- India AI Impact Summit: February 2026; debated India's global AI governance posture
- Paris AI Action Summit: February 2025; India signed the "Inclusive and Sustainable AI" declaration
- Aadhaar enrolments: 1.4 billion; UPI transactions FY2024: ~100–130 billion; DigiLocker users: 250+ million
- G7 Hiroshima AI Process (2023): voluntary code of conduct for advanced AI developers
- India's Bhashini mission: AI translation tools for India's 22 scheduled languages
- India ONDC: open protocol e-commerce network; as of 2024, ~80,000+ sellers onboarded
- Global AI compute concentration: top 10 AI companies control majority of GPU training capacity; all based in US, EU, or China