What Happened
- The India AI Impact Summit 2026 was held at Bharat Mandapam, New Delhi, from February 16-21, 2026, organised by the Ministry of Electronics and Information Technology (MeitY) under the IndiaAI Mission.
- The summit concluded with the New Delhi Declaration, endorsed by 88 countries and international organisations as of February 21, rising to 91 by February 24 — a broader endorsement than previous summits achieved.
- However, the Declaration was criticised for falling short of the binding commitments expected: it emphasised voluntary commitments, data sharing on real-world AI use, and improving AI for under-represented languages, but made no mention of coordinating government action to address AI risks — a retreat from the language of prior summits.
- The summit was the fourth in the global AI safety/governance summit series (Bletchley 2023 → Seoul 2024 → Paris 2025 → New Delhi 2026) and the first hosted by a Global South nation.
- Geopolitical fragmentation — particularly US-China competition over AI dominance and disagreements over binding safety mandates — prevented consensus on substantive governance commitments.
- Amnesty International and other civil society groups criticised the summit for failing to deliver binding human rights safeguards for AI deployment.
Static Topic Bridges
The Global AI Summit Series: Bletchley to New Delhi
The international AI governance summit process was initiated at Bletchley Park, UK in November 2023, convened by the British government with deliberately limited participation. Each subsequent summit has broadened the agenda and expanded the geographic footprint. The Seoul Summit (2024, co-hosted by UK and South Korea) introduced innovation and equitable access as themes alongside safety. The Paris AI Action Summit (2025, co-chaired by France and India) shifted toward an "open, multi-stakeholder" approach with five thematic tracks. The India summit in 2026 — the first in the Global South — framed its goal as "demonstrable impact", emphasising practical AI applications in health, finance, and agriculture for developing countries.
- Bletchley Park Summit (UK): November 2023; first summit; focused on frontier AI safety risks; limited participation
- AI Seoul Summit (South Korea + UK): 2024; added innovation and inclusion themes
- AI Action Summit (Paris, France + India): February 2025; five tracks including public interest AI, future of work, trust, governance; co-chaired by France and India
- India AI Impact Summit (New Delhi): February 16-21, 2026; hosted by MeitY/IndiaAI Mission; first Global South host
- New Delhi Declaration: Endorsed by 88→91 countries (as of Feb 24, 2026)
- Organising body: IndiaAI Mission under MeitY
Connection to this news: India's role has grown from co-chair of Paris 2025 to solo host of 2026, reflecting its aspiration to shape global AI governance norms from a Global South perspective.
AI Governance: Key Conceptual Frameworks
AI governance encompasses the rules, institutions, norms, and technical standards that govern the development, deployment, and use of artificial intelligence. Three broad governance approaches dominate current international debates: (1) the safety-first approach (UK/EU model), emphasising binding risk assessments, frontier AI oversight, and human rights safeguards; (2) the innovation-first approach (US under Trump-era policy, tech industry preference), emphasising voluntary commitments and market-led development; and (3) the development-oriented approach (Global South, India's articulation), emphasising AI for public good, multilingualism, and equitable access rather than primarily regulating risks. These three orientations are increasingly difficult to reconcile in a single international declaration.
- EU AI Act (2024): First binding horizontal AI law; risk-based classification (unacceptable, high, limited, minimal risk)
- US approach (Biden era): Executive Order on Safe, Secure, and Trustworthy AI (October 2023); Biden-era emphasis on safety; Trump (2025) revoked Biden's AI EO, emphasising innovation
- UK approach: Voluntary cross-sector principles; AI Safety Institute (Bletchley-launched)
- India's approach: IndiaAI Mission; AI for social good; emphasis on bridging digital divide, multilingual AI, inclusive deployment
- UN Secretary-General's AI Advisory Body: Recommended International AI Safety Institute and global AI governance treaty
Connection to this news: The divergence between these three approaches explains why the New Delhi Declaration remained aspirational rather than binding — geopolitical fragmentation made a stronger consensus impossible.
IndiaAI Mission and India's Domestic AI Policy
India's AI strategy is anchored in the IndiaAI Mission, launched in March 2024 with a ₹10,371.92 crore outlay. The mission has seven pillars: IndiaAI Compute Capacity (sovereign GPU infrastructure), IndiaAI Innovation Centre, IndiaAI Datasets Platform, IndiaAI Application Development Initiative, IndiaAI FutureSkills, IndiaAI Startup Financing, and Safe & Trusted AI. The compute backbone involves 10,000+ GPU clusters being established through public-private partnerships. India's AI governance approach emphasises responsible AI without creating regulatory barriers to innovation — a position that placed it between the EU's strict regulatory model and the US's laissez-faire approach during the summit negotiations.
- IndiaAI Mission: Launched March 2024; ₹10,371.92 crore outlay
- Seven pillars: Compute, Innovation Centre, Datasets Platform, App Development, FutureSkills, Startup Financing, Safe & Trusted AI
- GPU target: 10,000+ GPUs through public-private partnerships
- Key ministerial oversight: MeitY (Minister Ashwini Vaishnaw)
- India's AI governance philosophy: Responsible yet innovation-enabling; AI for public good and multilingual access
Connection to this news: India leveraged the summit to showcase the IndiaAI Mission's achievements and to position itself as a credible bridge between the developed world's AI safety concerns and the Global South's demand for equitable AI access.
Risks of Frontier AI and International Safety Concerns
Frontier AI refers to the most capable AI systems at or near the technological cutting edge. Key risks identified in international discussions include: catastrophic misuse (biological, chemical, cyber weapons enabled by AI); mass disinformation and democratic manipulation; autonomous weapons systems without meaningful human control; and systemic labour displacement. The Bletchley Declaration (2023) was the first international agreement to acknowledge that frontier AI may pose existential or catastrophic risks. Subsequent summits have progressively softened this language under pressure from innovation-focused governments and industry lobbies.
- Frontier AI risks: Catastrophic misuse; disinformation; autonomous weapons (LAWS); labour displacement; loss of human oversight
- Bletchley Declaration (2023): First international acknowledgment of frontier AI existential risks
- AI Safety Institutes: UK (AISI), US (USAISI), Japan, South Korea, Singapore — established post-Bletchley
- Voluntary "Frontier Safety Commitments" (Paris 2025): Industry pledges to share safety information with governments
- Civil society criticism of New Delhi Declaration: No binding human rights safeguards; failed to rein in state misuse of AI
Connection to this news: The New Delhi Declaration's retreat from risk-focused language reflects the tension between India's desire to serve as a neutral host and the impossibility of achieving substantive AI safety commitments in a fractured geopolitical environment.
Key Facts & Data
- Summit dates: February 16-21, 2026, Bharat Mandapam, New Delhi
- Organiser: MeitY under IndiaAI Mission
- New Delhi Declaration: Endorsed by 88 countries (Feb 21), rising to 91 (Feb 24)
- Summit series: Bletchley (Nov 2023) → Seoul (2024) → Paris (Feb 2025) → New Delhi (Feb 2026)
- India is the first Global South nation to host the global AI summit
- IndiaAI Mission outlay: ₹10,371.92 crore (launched March 2024)
- India's GPU target: 10,000+ GPUs through public-private partnerships
- Criticism: No binding human rights safeguards; voluntary commitments only; no mention of coordinating on AI risk
- EU AI Act (2024): First binding AI law globally; risk-based classification
- Over 35,000 registrations from 100+ countries for the summit
- Key side event: India joins Pax Silica (February 20, 2026)