Every AI Policy Ever · open in the interactive register

Published (non-binding, current)Non-binding · Declaration / statement

AI Impact Summit Declaration

Record INT-02

Issued by
Participants of the India AI Impact Summit 2026, New Delhi (hosted by the Government of India)
Jurisdiction
AI Impact Summit (New Delhi, 2026)
Level · region
International · Global
Published / adopted
2026-02
Last updated
2026-03-05
Language
English

Summary

Declaration of the AI Impact Summit in New Delhi, held after the Bletchley, Seoul and Paris summits. It calls for international cooperation along seven pillars: human capital, access for social empowerment, trustworthiness of AI systems, energy efficiency, AI in science, democratising AI resources, and AI for economic growth and social good. Its secure and trusted AI section calls secure, trustworthy and robust AI foundational to trust and recognises security in AI systems, industry-led voluntary measures, technical solutions and policy frameworks across the AI lifecycle. It notes the voluntary Trusted AI Commons, a platform of technical resources, tools, benchmarks and best practices.

Safety topics

international cooperationevaluation & testing

Status notes

The declaration is headed 'AI Impact Summit Declaration, New Delhi (February 18 - 19, 2026)', and its preamble says the participants gathered in New Delhi on 19 February 2026; the European Commission announced on 20 February 2026 that the EU would endorse it, and the Australian government gives the summit as 19 and 20 February 2026, so only the month is given. The Commission calls it the Leaders' Declaration at the AI Impact Summit, the Australian government the India AI Impact Summit Declaration. The endorsement list published with the text names 89 endorsers: 87 countries, among them the USA, China, India, Russia and the UK, plus the EU and IFAD. Under its pillars the declaration takes note of voluntary deliverables, among them the Charter for the Democratic Diffusion of AI, the Global AI Impact Commons, the Trusted AI Commons and its guidance note, the International Network of AI for Science Institutions and the Voluntary Guiding Principles on Resilient, Innovative, and Efficient Artificial Intelligence.

Sources

Open the official source https://www.pib.gov.in/PressReleasePage.aspx?PRID=2231208

Secondary source: https://eoibelgrade.gov.in/listview/Njk1

Local copy

Partial (secondary source only). The official source could not be downloaded (DNS lookup failed, so the site could not be reached from this machine (try another network, or save the document by hand). robots.txt could not be checked, which RFC 9309 treats as 'do not crawl'); only the secondary source was saved. Last attempt 2026-10-17 16:29:29 UTC

Secondary source (third-party copy, not published) archive/INT-02/2026-10-17/secondary.html
Retrieved 2026-10-17 16:29:34 UTC
SHA-256 8108fc02abcb00b1824474abbecff0d6a5b11b60be2c4a0a05d5376b51e226cf

Verification

A
The instrument's own official text, or an official register, was stored and confirms the title, date and status. checked 2026-10-17 · official-url-fetched
pib.gov.in and impact.indiaai.gov.in, the record's sources, could not be reached from the archiver's network on 11, 16 and 17 Oct 2026 (robots.txt unreachable, then DNS failures), and other Indian government hosts (mea.gov.in, negd.gov.in) were unreachable or geo-blocked. Stored and read in this run: the declaration's full text with its endorsement list as published by the Embassy of India in Belgrade (eoibelgrade.gov.in, Government of India; proposed as secondary_url), which confirms the title, the dateline 18-19 February 2026, the pillars, the secure-and-trusted-AI section and the voluntary, non-binding character; the European Commission's press release of 20 February 2026 (digital-strategy.ec.europa.eu, alt-official) on the EU's endorsement; and the Australian Department of Industry, Science and Resources' news item of 23 February 2026 (industry.gov.au, alt-official), which links the text on the Ministry of External Affairs website (https://www.mea.gov.in/bilateral-documents.htm?dtl%2F40809%2F, unreachable from the archiver's network). Version 1's title, 'New Delhi Declaration on AI Impact', is not the heading of the stored text; the declaration's own heading is used now. Version 1's 92 endorsers as of 5 March 2026, the names of the three Sutras, and its companion documents (a Guidance Note on AI Governance, 'AI and Child Safety', the New Delhi Frontier AI Impact Commitments) rest on the PIB releases, which are not stored.
Second check: Independent check by Claude Code (claude-sonnet-5-5), 2026-10-17: 9 field(s) supported.

History

  1. 2026-10-18related: [{"slug": "paris-ai-action-summit-statement", "id": "INT-01"}, {"slug": "interna → [{"slug": "paris-ai-action-summit-statement", "id": "INT-01"}, {"slug": "interna — two-way link added: BRICS-02, IN-02 link(s) to INT-022026-10-18-fix-related-links · Claude Code (claude-opus-5-5)
  2. 2026-10-17notes: The summit's high-level segment was held in New Delhi in Feb 2026; PIB gives 18- → The declaration is headed 'AI Impact Summit Declaration, New Delhi (February 18 — rewritten from the stored text and the EU and Australian items: the dateline and dates, the other names, the 89 endorsers of the published list and the deliverables the text names; claims resting on the unstored PIB releases removed (see open_questions)2026-10-17-recheck-international-2 · Claude Code (claude-opus-5-5)
  3. 2026-10-17safety_topics: ["international cooperation", "evaluation & testing", "research priorities"] → ["international cooperation", "evaluation & testing"] — 'research priorities' dropped: the text's science pillar is about AI for science, not safety research; 'international cooperation' (preamble) and 'evaluation & testing' (benchmarks in the Trusted AI Commons) stay2026-10-17-recheck-international-2 · Claude Code (claude-opus-5-5)
  4. 2026-10-17summary: This outcome of the fourth summit in the series is built on three 'Sutras' (Peop → Declaration of the AI Impact Summit in New Delhi, held after the Bletchley, Seou — rewritten from the stored text: the three Sutras are not named in it, 'potential security risks' is 'potential security aspects' in the text, and the Trusted AI Commons is described as the text describes it (technical resources, tools, benchmarks and best practices, not 'safety tools'); the preceding summits from the Australian government's item2026-10-17-recheck-international-2 · Claude Code (claude-opus-5-5)
  5. 2026-10-17secondary_url: https://impact.indiaai.gov.in/outcome-resources → https://eoibelgrade.gov.in/listview/Njk1 — the declaration's full text with its endorsement list, published by the Embassy of India in Belgrade (Government of India), stored in this run; the old secondary URL (impact.indiaai.gov.in outcome list) is a list page that does not resolve from the archiver's network2026-10-17-recheck-international-2 · Claude Code (claude-opus-5-5)
  6. 2026-10-17title: New Delhi Declaration on AI Impact → AI Impact Summit Declaration — the declaration's own heading on the Government of India's published text is 'AI Impact Summit Declaration' (the EU and Australia name it after the summit too); version 1's 'New Delhi Declaration on AI Impact' came from the PIB release, which is not stored; 'New Delhi Declaration' stays as short_title2026-10-17-recheck-international-2 · Claude Code (claude-opus-5-5)

Related records

Paris AI Action Summit Statement AI Action Summit (Paris, 2025) · Published (non-binding, current)

International AI Safety Report 2026 International AI Safety Report (AI Safety Summit process) · Published (non-binding, current)

BRICS New Delhi Declaration 2026 (AI paragraphs) BRICS · Published (non-binding, current)

India AI Governance Guidelines India · Published (non-binding, current)

Citation

Participants of the India AI Impact Summit 2026, New Delhi (hosted by the Government of India). AI Impact Summit Declaration. 2026-02. https://www.pib.gov.in/PressReleasePage.aspx?PRID=2231208. Status: Published (non-binding, current); verified 2026-10-17 (level A). Local copy archive/INT-02/2026-10-17/secondary.html retrieved 2026-10-17 16:29:34 UTC, SHA-256 8108fc02abcb00b1824474abbecff0d6a5b11b60be2c4a0a05d5376b51e226cf. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record INT-02.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.