Last reviewed
May 18, 2026
Sources
No citations yet
Review status
Needs citations
Revision
v1 · 6,594 words
Add missing citations, update stale details, or suggest a clearer explanation.
The Paris AI Action Summit (French: Sommet pour l'action sur l'IA) was the third major international summit on artificial intelligence in the so-called "AI Safety Summit" series, held on 10–11 February 2025 at the Grand Palais in Paris and co-chaired by French President Emmanuel Macron and Indian Prime Minister Narendra Modi.[^1][^2] Around 100 countries and more than 1,000 participants — heads of state and government, ministers, leaders of international organisations, chief executives of major artificial-intelligence companies, researchers and civil-society representatives — attended the two-day plenary, which was preceded by almost a week of side events under the umbrella "AI Action Week" running from 6 February.[^1][^3][^4]
The summit produced a non-binding political declaration titled Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet, signed by around sixty countries together with the European Union and the African Union Commission.[^5][^6][^7] Crucially, both the United States and the United Kingdom declined to sign the declaration, with the U.S. citing concerns about "excessive regulation" and the U.K. citing insufficient "practical clarity" on global governance and national-security risks.[^8][^9][^10]
The Paris summit is widely regarded as a turning point in the global AI policy conversation. The 2023 Bletchley AI Safety Summit in the United Kingdom and the May 2024 AI Seoul Summit in South Korea had framed advanced AI primarily through the lens of safety and catastrophic risk; Paris dropped the word "safety" from the meeting's name in favour of "action" and oriented its agenda around investment, public-interest AI, sustainability and the governance of artificial intelligence rather than around frontier risk mitigation.[^11][^12][^13] U.S. Vice-President J.D. Vance used his keynote address on 11 February to deliver a forceful critique of European AI regulation and of cooperation with "authoritarian regimes" on AI infrastructure, signalling a sharp turn in U.S. policy under the second Trump administration.[^14][^15][^16] French President Macron used the same platform to unveil a €109 billion private-investment plan for AI data-centre build-out in France and to host European Commission President Ursula von der Leyen, who announced the InvestAI initiative to mobilise €200 billion across the European Union.[^17][^18][^19]
The next event in the series, the India AI Impact Summit, was announced from the Paris stage by Prime Minister Modi and ultimately held in New Delhi in February 2026.[^20][^21]
The Paris AI Action Summit was the third instalment of a multinational summit series initiated by the United Kingdom in November 2023. The first summit, held at the historic Bletchley Park codebreaking site, produced the Bletchley Declaration — a joint statement signed by 28 countries and the European Union (29 signatories in total) acknowledging the "potential for serious, even catastrophic, harm" from frontier AI systems and committing signatories to a shared scientific understanding of frontier risk.[^11][^22] The Bletchley summit also commissioned what became the International AI Safety Report under the chairmanship of Turing Award laureate Yoshua Bengio.[^23]
A follow-up "interim" AI Safety Summit, the AI Seoul Summit, was co-hosted by the United Kingdom and the Republic of Korea on 21–22 May 2024.[^11][^24] Seoul produced the Seoul Declaration and the Frontier AI Safety Commitments, in which 16 leading AI companies — including OpenAI, Anthropic, Google DeepMind, Microsoft and Mistral AI — pledged to publish frontier-safety frameworks and to pause development of systems whose risks could not be sufficiently mitigated.[^11][^24] The Seoul summit further codified the network of national AI Safety Institutes that the U.K. and U.S. had launched at Bletchley.[^11][^24]
When France was nominated to host the third summit, it announced in mid-2024 that it would rebrand the event as the AI Action Summit, deliberately downplaying the singular focus on existential or catastrophic risk in favour of a broader agenda framed around the public interest, sustainability, inclusive access to AI and the labour market.[^11][^13][^25] French AI envoy Anne Bouverot, appointed by Macron to organise the summit, told journalists that fears of human-level or superintelligent AI taking control should be regarded as "science fiction" and not as the primary problem of multilateral AI policy.[^13][^25] That framing shaped what observers — both supportive and critical — described as a "vibe shift" in the international AI policy conversation between Seoul in May 2024 and Paris in February 2025.[^11][^12][^13]
Two further developments shaped the immediate run-up to the Paris summit. On 21 January 2025, U.S. President Donald Trump, less than 24 hours after his second inauguration, announced the Stargate Project — a venture promising up to US$500 billion in private-sector U.S. AI infrastructure investment, anchored by OpenAI, Oracle, SoftBank and the Emirati sovereign wealth fund MGX.[^26][^27] One week later, the Chinese startup DeepSeek published DeepSeek-R1, a high-performing reasoning model trained for a fraction of the cost of comparable U.S. frontier models, triggering a sharp sell-off in U.S. semiconductor stocks and intensifying the U.S.–China AI race.[^26][^27][^28] These events meant that by the time delegates arrived in Paris, the geopolitical centre of gravity around AI had moved decisively toward great-power competition.[^26][^28]
The French presidency framed the Paris summit as the culmination of an entire "AI Action Week" running from 6 to 11 February 2025, with thematic side events spread across multiple Paris venues.[^3][^29]
On 6–7 February, the AI, Science and Society conference was hosted by the Institut Polytechnique de Paris (IP Paris) at the École Polytechnique campus in Palaiseau, designed to provide diplomats with a scientific overview to underpin the summit discussions.[^3][^29] Speakers included Meta chief scientist Yann LeCun, Max Planck Institute director Bernhard Schölkopf, Stanford economist Erik Brynjolfsson and others, with sessions on foundation models, generative AI, AI for science and trustworthy AI.[^29]
On 8–9 February, public-facing "culture days" were held at the Bibliothèque nationale de France and the Conciergerie, focused on art, creative industries and the implications of AI for cultural production.[^3] On 9 February, President Macron pre-empted criticism of the summit's marketing tone by posting an Instagram clip of AI-generated deepfake videos of himself, telling the public that AI could "change healthcare, energy, life in our society."[^25][^30]
The plenary at the Grand Palais ran on 10–11 February. On the evening of 10 February, Macron hosted a state dinner at the Élysée Palace bringing together heads of state and government, leaders of international organisations and chief executives of major AI companies.[^1][^31] On 11 February, Macron and Modi co-chaired the formal plenary session, after which over 100 ancillary events — including a business-focused day at the Station F start-up campus — were held across Paris.[^1][^3]
The publication of the first International AI Safety Report on 29 January 2025, designed expressly to inform the Paris discussions, was an important precursor (see International AI Safety Report section below).[^23][^32]
The French presidency reported that the plenary brought together roughly 100 countries, approximately 90 government representatives at head-of-state, ministerial or senior-administrative level, around 20 representatives of international organisations and senior representatives of major businesses, alongside scientists, academics and civil-society organisations — for a total of more than 1,000 participants.[^1][^31][^33]
Among heads of state and senior government leaders confirmed in attendance were:
The senior private-sector roster included Sam Altman (CEO, OpenAI), Sundar Pichai (CEO, Google), Demis Hassabis (CEO, Google DeepMind), Brad Smith (President, Microsoft), Dario Amodei (CEO, Anthropic), Jamie Dimon (CEO, JPMorgan Chase), Nick Clegg (then-President of Global Affairs, Meta), Lisa Su (CEO, AMD) and Arthur Mensch (co-founder and CEO of Mistral AI), among many others.[^31][^39][^40] Sundar Pichai delivered remarks at the Grand Palais, while Sam Altman and Dario Amodei sat in the audience for Vice-President Vance's address.[^39][^40]
Notable for absences: U.S. President Donald Trump and U.K. Prime Minister Keir Starmer did not attend in person, although both governments were represented.[^9][^10][^14]
The French presidency organised the summit programme around five working themes that had been the subject of preparatory contact groups in the months before February 2025:[^3][^29][^33]
Critics noted that compared with Bletchley and Seoul, safety occupied a far smaller share of the agenda — relegated to one of five working strands rather than being the organising principle of the summit.[^11][^12][^13]
The opening day at the Grand Palais featured a multi-stakeholder forum bringing together delegations, civil-society organisations, researchers and business leaders for plenary and parallel discussions across the five working themes.[^1][^3] Macron used Day 1 to announce that France had secured around €109 billion ($112 billion) in private commitments for AI infrastructure in France, an announcement timed to coincide with the opening day of the summit.[^17][^41][^42] The announcement was made before the opening plenary in an interview broadcast on French television.[^17][^41][^42]
In the evening, Macron hosted a state dinner at the Élysée Palace.[^1][^31] At the dinner and surrounding receptions, AI company executives mingled with heads of state, and Mistral AI co-founder Arthur Mensch announced a series of strategic partnerships including with the French telecoms group Orange, the Italian carmaker Stellantis and the French waste-and-water utility Veolia, plus a partnership with German defence-technology start-up Helsing.[^39][^43][^44]
The formal head-of-state plenary on 11 February began with Macron's address and an opening address from Modi as co-chair.[^2][^34][^45] The plenary then heard a sequence of head-of-state and head-of-government interventions, ministerial statements and addresses from the leaders of international organisations.[^1][^33]
Macron used his keynote to argue that Europe must be "in the race" for AI development and that France, with its decarbonised nuclear-powered electricity grid, was well placed to host the energy-intensive computation required for frontier AI training.[^17][^41][^42] He pledged that France would dedicate one gigawatt of nuclear capacity to AI training by the end of 2026 and characterised the €109 billion private-investment package as "exactly the equivalent for France of what the United States announced with Stargate" — at $500 billion — given France's roughly five-times-smaller population.[^17][^41][^42] Macron also defended Europe's regulatory framework, arguing that the EU AI Act could be simplified but not abandoned, and that European rules need not preclude European innovation.[^17][^25]
Modi, as co-chair, delivered the opening address of the 11 February plenary at Macron's invitation. He argued that "AI is writing the code for humanity in this century" and proposed a five-point agenda: (i) pool global resources and talent for AI innovation; (ii) develop open-source AI systems to foster trust and transparency; (iii) create high-quality, unbiased datasets for responsible AI training; (iv) democratise AI to develop people-centric applications; and (v) address cybersecurity risks, disinformation and deepfakes.[^34][^45] He emphasised the need to "democratise technology" and ensure "access to all, especially in the Global South," noted that India was building its own large language model tailored to its linguistic and cultural diversity, and announced from the Paris stage that India would host the next AI summit.[^34][^45][^46]
European Commission President Ursula von der Leyen used her plenary intervention to announce InvestAI, an EU initiative to mobilise up to €200 billion of investment in artificial intelligence across the Union, including a €20 billion European fund for "AI gigafactories" — large computing facilities each equipped with around 100,000 next-generation AI accelerators for training frontier models.[^18][^19][^47] Von der Leyen described the partnership as a "CERN for AI" and argued that the scheme would democratise compute access for European scientists and small companies as well as large incumbents.[^18][^19] She also acknowledged a parallel private-sector "EU AI Champions" initiative led by venture-capital firm General Catalyst, which had committed €150 billion of private investment over five years from a coalition of more than 60 European companies and investors — including Mistral AI, Helsing, Spotify and Deutsche Bank.[^48][^49]
Google CEO Sundar Pichai used his stage time to argue that "AI is a once in a lifetime technology" and that "the biggest risk could be missing out" on its benefits, a phrasing widely picked up by observers as emblematic of the summit's shift in tone from caution to acceleration.[^40][^50] Pichai described Google's decade-plus of AI investment, highlighted breakthroughs including Demis Hassabis's Nobel-prize-winning work on AlphaFold (now used by 2.5 million researchers in more than 190 countries), and announced a Global AI Opportunity Fund targeting 20,000 trainees across 24 countries in Europe.[^40][^50] He called for regulatory alignment "across countries" to prevent a fragmented international rulebook.[^40][^50]
The most consequential intervention of Day 2 was the keynote address by U.S. Vice-President J.D. Vance, the first foreign trip of his tenure following his inauguration on 20 January 2025.[^14][^15][^16][^26] Vance used the Paris platform to deliver what U.S. and European observers described as a fundamental reset of the United States' international position on AI policy.[^14][^15][^16][^51]
Vance opened by declaring that "I'm not here this morning to talk about AI safety, which was the title of the conference a couple of years ago. I'm here to talk about AI opportunity."[^14][^15][^52] He went on to lay out four priorities of the second Trump administration's AI policy:[^14][^15][^16]
Vance also issued a pointed warning against partnerships with "authoritarian regimes" on AI infrastructure — widely interpreted as a reference to China — arguing that allies should resist the temptation of accepting "cheap" or subsidised AI tools that could ultimately leave them dependent.[^14][^15][^51] He told the audience that "the AI future is not going to be won by hand-wringing about safety" and that "the AI economy will primarily depend on and transform the world of atoms" — power plants, semiconductor fabs and physical infrastructure.[^14][^15][^52]
Vance's address marked a sharp public departure from the AI-safety language his predecessor Vice-President Kamala Harris had used at Bletchley in November 2023, and it provided the immediate context for the U.S. decision not to sign the Paris declaration.[^14][^15][^51][^54] Sam Altman, who was in the audience, publicly applauded the speech.[^55] Sundar Pichai sat on stage behind Vance during the address.[^39][^40]
The United Kingdom — host of the original 2023 Bletchley summit — declined to sign the Paris declaration alongside the United States. A spokesperson for Prime Minister Keir Starmer's office stated that the U.K. government would "only ever sign up to initiatives that are in the U.K.'s national interests" and that the declaration "didn't provide enough practical clarity on global governance" or sufficiently address national-security concerns.[^9][^10][^38] Asked directly whether the decision was made to align with the Trump administration, U.K. officials denied this characterisation.[^9][^38]
Technology Secretary Peter Kyle represented the U.K. in Paris. The decision was widely criticised in the U.K., with the fact-checking organisation Full Fact and AI-policy commentators arguing that London was undermining its credibility as a leader on safe AI — a leadership position the U.K. had cultivated since 2023 through the Bletchley Declaration, the establishment of the U.K. AI Safety Institute (later renamed AI Security Institute) and the launch of the International AI Safety Report.[^10][^38][^54]
The summit's principal multilateral output was the Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet, a short non-binding political declaration agreed by around sixty countries plus the European Union and the African Union Commission.[^5][^6][^7][^56] The Élysée Palace published the official English text on 11 February 2025.[^5]
The statement articulated six main priorities:[^5][^6]
The text also referenced the launch of a Public Interest AI Platform and Incubator, an AI and Energy discussion process, and an observatories network on AI's labour-market impacts.[^5][^6][^56]
Wikipedia and the official French government channel list the signatories as the European Union, the African Union Commission, and (according to varying tallies) between 58 and 61 individual countries.[^5][^6][^7] News reports cited the figure as "around 60" or "more than 60 countries, plus the EU and the African Union," with the discrepancy between 58, 60 and 61 reflecting different ways of counting late additions and the EU/AU as parties rather than countries.[^7][^57][^58]
The known signatory countries include:[^5][^6]
Armenia, Australia, Austria, Belgium, Brazil, Bulgaria, Cambodia, Canada, Chile, China, Croatia, Cyprus, Czechia, Denmark, Djibouti, Estonia, Finland, France, Germany, Greece, Hungary, India, Indonesia, Ireland, Italy, Japan, Kazakhstan, Kenya, Latvia, Lithuania, Luxembourg, Malta, Mexico, Monaco, Morocco, Netherlands, New Zealand, Nigeria, Norway, Poland, Portugal, Republic of Korea, Romania, Rwanda, Senegal, Serbia, Singapore, Slovakia, Slovenia, South Africa, Spain, Sweden, Switzerland, Thailand, Ukraine, United Arab Emirates, Uruguay and the Holy See (Vatican), together with the European Union and the African Union Commission.
The most consequential non-signatories were the United States and the United Kingdom.[^7][^8][^9][^10][^57] Russia and several other major states were not signatories either, but the U.S. and U.K. abstention attracted by far the most attention given their previous central role in the AI summit series.[^11][^54] China's signature was widely noted, given that China had also signed the Bletchley Declaration in 2023 and was sending a vice-premier rather than President Xi himself to Paris.[^4][^36][^58]
A defining message of the Paris summit was the scale of new European AI-infrastructure investment announced. French President Macron, in interviews broadcast on the morning of 10 February and at the plenary the following day, announced €109 billion (≈US$112 billion) of private commitments to AI infrastructure in France.[^17][^41][^42]
According to figures collated by news outlets covering the announcement, the largest pledges included:[^17][^41][^42]
Macron explicitly framed the package as France's structural answer to the United States' Stargate Project, announced three weeks earlier, arguing that on a per-capita basis the French commitment was proportionately comparable to the $500 billion Stargate figure.[^17][^41][^42]
European Commission President Ursula von der Leyen complemented Macron's announcement on 11 February with the InvestAI initiative — a public-private package designed to mobilise €200 billion for AI across the European Union.[^18][^19][^47] The headline element was a new €20 billion European fund for AI gigafactories, with the goal of building four large-scale public-private compute facilities each equipped with roughly 100,000 next-generation AI chips, intended to be operational over 2027–2028.[^18][^19][^47] InvestAI is funded by the EU budget (Digital Europe, Horizon Europe, InvestEU), with member-state top-ups and a public first-loss tranche designed to de-risk private investment.[^18][^19] Von der Leyen described the InvestAI gigafactories programme as analogous to "CERN for AI."[^18][^19]
The privately led EU AI Champions Initiative, organised by the U.S. venture-capital firm General Catalyst and announced alongside the summit, added a further €150 billion of private investment commitment from more than 60 European companies and investors over five years.[^48][^49] Combined, the European headline figures totalled around €200 billion of mobilised investment under InvestAI plus the €109 billion in France-specific private commitments, with significant overlap between the two.[^17][^18][^48][^49]
The Paris summit also produced a series of multistakeholder initiatives launched alongside (rather than within) the official declaration:[^25][^59][^60]
The Paris summit was the formal venue for the first edition of the International AI Safety Report, an inter-governmental synthesis of the scientific literature on the capabilities and risks of general-purpose AI, chaired by Université de Montréal professor and 2018 Turing Award co-recipient Yoshua Bengio.[^23][^32][^63] The report had been commissioned at the 2023 Bletchley summit and was published on 29 January 2025, two weeks ahead of the Paris meeting.[^23][^32][^63]
The first edition was the work of 96 AI experts, supported by an Expert Advisory Panel drawn from 30 countries together with the United Nations, European Union and OECD.[^23][^32][^63] The report did not make policy recommendations; instead, it synthesised the state of scientific evidence on three categories of risk:[^23][^32]
The report concluded that AI capabilities had advanced "faster than many experts anticipated" since the Bletchley summit, that the evidence for several risks had "grown substantially," and that risk-management techniques were "improving but insufficient."[^23][^63] Bengio publicly warned in Paris that a "frantic race" by major tech companies to develop ever more powerful AI could have "harmful" effects, echoing similar warnings made over the previous two years by him and by Geoffrey Hinton.[^64][^65]
Bengio's report was presented to delegations at Paris but, unlike the Bletchley and Seoul declarations, was not directly referenced in operative paragraphs of the Statement on Inclusive and Sustainable AI — a point critics highlighted as evidence that safety had been demoted from the centre of the agenda.[^11][^13][^25][^66]
The Paris summit elicited an unusually polarised reaction from leading figures in the AI industry and the AI research community.
On 11 February, Anthropic published a statement under the name of CEO Dario Amodei describing the Paris summit as a "missed opportunity."[^67][^68] Amodei argued that the international community had failed to give sufficient priority to three issues: (i) ensuring democracies retain the lead in advanced AI development relative to authoritarian states; (ii) managing security risks, including the danger that increasingly capable systems might be misused by non-state actors against critical infrastructure or in loss-of-control scenarios; and (iii) preparing for the economic disruption that frontier AI is likely to produce.[^67][^68]
Anthropic's pre-summit warning included Amodei's now widely cited timeline estimate that by 2026 or 2027, and "almost certainly no later than 2030," AI systems could have capabilities "akin to an entirely new state populated by highly intelligent people appearing on the global stage" — a formulation he had developed in his October 2024 essay Machines of Loving Grace.[^67][^68] Amodei wrote that "Greater focus and urgency is needed" on these issues at future AI summits.[^67][^68]
OpenAI CEO Sam Altman, by contrast, told Axios on the margins of the summit that there was a "very different energy" in Paris compared with previous AI-safety meetings, with delegates "really excited about" the technology rather than fixated on its risks.[^39][^55] On the eve of the summit, Altman published an essay titled Three Observations on his personal blog, in which he argued that model performance continues to scale with compute, that costs continue to fall as scale grows, and that AI's economic impact will be "gigantic."[^55] Altman publicly applauded J.D. Vance's address.[^55]
Sundar Pichai's "biggest risk could be missing out" framing — delivered from the Grand Palais stage on 11 February — was widely interpreted as a marker of the new industry consensus around acceleration.[^40][^50] Demis Hassabis, who attended the summit as CEO of Google DeepMind and was specifically recognised by Pichai for his 2024 Nobel Prize in Chemistry for AlphaFold, joined Pichai in framing AI as a transformative scientific platform.[^40][^50]
Yoshua Bengio used his Paris appearances to argue that the report he had chaired showed "the evidence for several risks has grown substantially" and that policymakers face an "evidence dilemma" — having to act on incomplete data rather than wait for definitive proof that may come too late.[^23][^64] Fellow Geoffrey Hinton, although not delivering keynote remarks at the Grand Palais plenary, had publicly warned in the weeks before Paris that the global race to ever-more-powerful AI was dangerous, and his arguments framed civil-society commentary at the summit.[^65][^66]
Max Tegmark, MIT physicist and founder of the Future of Life Institute, characterised the Paris summit as "a tremendous missed opportunity" and described the declaration's omission of any meaningful safety commitments as "a recipe for disaster."[^66][^69] The Future of Life Institute had publicly called on countries not to sign the declaration on grounds that it was excessively vague.[^66][^69] David Leslie of the Alan Turing Institute argued that "the communique doesn't adequately speak to the sort of real-world risks and harms" from AI.[^66][^69] Jessica Galissaire of Renaissance Numérique said terms such as "sustainable" and "inclusive" became "devoid of any meaning" without shared definitions.[^66][^69]
The Paris summit's reception broke along three broad fault lines.[^11][^13][^25][^54][^70]
On the favourable side, the French presidency, the European Commission and many supportive commentators argued that the summit had successfully:[^17][^18][^19][^25][^71]
On the critical side, civil-society organisations, much of the AI-safety research community and many media observers argued that the summit had:[^11][^13][^25][^54][^66][^69][^70]
On the geopolitical fault line, commentators from think-tanks including CSIS, Chatham House and RUSI argued that the summit had crystallised a three-way split in international AI governance: a U.S. approach centred on innovation, hardware dominance and decoupling from China; a European approach centred on regulation, sovereignty and large-scale public-private investment; and a Chinese approach centred on inclusivity rhetoric, state-led governance and outreach to the Global South in the wake of DeepSeek's emergence.[^28][^36][^54][^58][^70] The Chinese vice-premier Zhang Guoqing told the summit that China was willing "to work with other countries to promote development, safeguard security, and share achievements in the field of artificial intelligence."[^36][^58]
By signing the declaration alongside China and most EU members while the U.S. and U.K. did not, India and the African Union Commission signalled an interest in non-aligned engagement on AI policy that would shape the build-up to the 2026 successor summit.[^46][^54][^58]
In his opening address on 11 February, Prime Minister Modi announced that India would host the next event in the summit series, formalising a decision that had been signposted in pre-summit diplomatic exchanges.[^34][^45][^46] The successor event, organised under the IndiaAI Mission of India's Ministry of Electronics and Information Technology, was renamed the AI Impact Summit to align with Modi's preference for "impact" over "safety" or "action" and was held at the Bharat Mandapam convention centre in New Delhi in February 2026.[^21][^46] It was the first summit in the series to be hosted by a Global South nation.[^21][^46]
Building on the Paris summit's themes, the New Delhi event placed particular emphasis on public-interest AI, on data infrastructure for the Global South, and on the labour-market and developmental implications of AI for emerging economies, and a second edition of the International AI Safety Report — also chaired by Bengio — was published on 3 February 2026 ahead of the New Delhi plenary.[^23][^46] Switzerland was expected to host a subsequent summit, tentatively scheduled in Geneva in 2027.[^21][^46]
The Paris AI Action Summit is widely regarded as the moment at which the multilateral AI conversation rotated away from a narrow safety-and-frontier-risk framing toward a broader political contest over regulation, investment, sovereignty and access to compute.[^11][^13][^25][^54][^70] By the time Paris closed, three blocs had defined themselves publicly: a deregulation-first United States led by Vice-President Vance, a regulation-with-investment European bloc led by Macron and von der Leyen, and an inclusive-development bloc centred on China, India and many Global South states.[^28][^36][^54][^58][^70] The summit also marked the public arrival of an explicitly "Action"-framed European AI policy of large-scale industrial investment — symbolised by InvestAI's €200 billion target and the announcement of AI gigafactories on the model of "CERN for AI."[^18][^19][^47]
Whether the Paris summit ultimately strengthened or weakened multilateral AI governance remained contested at the time of the New Delhi summit a year later. Supporters argued it had opened the conversation to dozens more countries and to substantive industrial and developmental questions. Critics — including a substantial fraction of the AI-safety research community — argued that the summit had quietly closed the brief window opened at Bletchley for global agreement on frontier-risk governance, even as the underlying capabilities of frontier AI systems continued to advance.[^11][^13][^54][^66][^67][^69]