PauseAI international protest wave around the Paris AI Action Summit (7–11 February 2025)

Between 7 and 11 February 2025, PauseAI staged a coordinated international protest wave across roughly twenty cities on five continents, timed to the AI Action Summit that French President Emmanuel Macron and Indian Prime Minister Narendra Modi co-chaired at the Grand Palais in Paris on 10–11 February. The wave's anchor demonstration was held in Paris by the Pause IA French chapter, with sibling actions in New York, Victoria, London, Berlin, Brussels, Zürich, Prague, Milan, Stockholm, Copenhagen, Oslo, Kristiansand, Trondheim, Kinshasa, Brazzaville, N'Djaména, Yaoundé, and Melbourne, and ran alongside the Pause IA-organised Reprendre le Contrôle civil-society conference at the Learning Planet Institute on 8 February. It is the moment at which the PauseAI international protests campaign shifted from a primarily Northern-hemisphere footprint to its first explicit African-continent presence, and at which the campaign's complaint that "Action" had displaced "Safety" in the Summit's stated programme moved from PauseAI's own pages onto French-language broadcast and into the Summit week's public record.

Context

The Summit was the third meeting in the international AI-summit series that had begun at Bletchley Park in November 2023 and continued with Seoul in May 2024. Where the Bletchley summit had been styled as an "AI Safety Summit" — with a declaration on coordinated work on frontier-AI risk as its signature output — and the Seoul summit had retained that framing, the Paris meeting was branded as an "AI Action Summit", with a five-track programme covering public-interest AI, the future of work, innovation and culture, trust in AI, and global AI governance. PauseAI's public case against the Summit's framing was that the rename was substantive rather than cosmetic: the safety-and-binding-rules direction of the Bletchley track had been displaced by a Summit programme heavier on capability deployment, public-interest infrastructure, and innovation policy than on the systemic-risk and treaty-architecture questions the campaign held to be the field's central problem.

In the weeks before the Summit, PauseAI launched a Change.org petition asking Summit organisers to allocate dedicated programme time to three demands: at least one session on creating global treaties to mitigate catastrophic AI risk; at least one session on establishing international AI safety enforcement bodies; and facilitation for participating nations to share population-protection best practices. The protest wave was framed as the public-mobilisation arm of that ask — a visible international demonstration that AI safety remained the issue summit-bound ministers were under public pressure to take up, regardless of the Summit's renamed programme structure.

What happened during the protest wave

The wave opened on Friday 7 February 2025 with the first sibling actions and ran through Tuesday 11 February, the Summit's closing day. PauseAI's master index entry gives a fifteen-plus-city count for the master record; the campaign's own retrospective names nineteen cities. The full action footprint, set out on the /2025-february campaign page, spans:

• France: Paris.
• Other Europe: London, Berlin, Brussels, Zürich, Prague, Milan, Stockholm, Copenhagen, Oslo, Kristiansand, Trondheim.
• North America: New York (United States), Victoria (Canada).
• Africa: Kinshasa (DR Congo), Brazzaville (Republic of Congo), N'Djaména (Chad), Yaoundé (Cameroon).
• Oceania: Melbourne (Australia).

The Paris anchor demonstration was organised by Pause IA's Antoine de Scorraille and Maxime Fournes and staged at the Place de la Bastille in the 4th arrondissement from 8:30 AM on Monday 10 February — the Summit's opening day at the Grand Palais — with thirty-one signups recorded on the chapter's Luma page. The chapter's stated frame for the host-city action was that French citizens would publicly demonstrate that they "would not remain silent in the face of AI-related threats" and would resist industry pressure to displace safety from the Summit's centre — the same framing the chapter's leadership, including Fournes, carried into French-language broadcast across the Summit week on Cause Commune's Parlez-moi d'IA episode #56.

Two days before the street action, on Saturday 8 February, Pause IA opened its Summit-week programme with the Reprendre le Contrôle — Forum des Solutions pour une IA Compatible avec l'Humanité conference at the Learning Planet Institute in Paris. The conference brought together roughly a hundred experts, researchers, and decision-makers and featured speakers including Lê Nguyên-Hoang (Calicarpa), Lou Welgryn (Data for Good), Arthur Grimonpont (Reporters Sans Frontières), Raja Chatila (Sorbonne Université), and Charbel-Raphaël Segerie (Centre pour la Sécurité de l'IA), with co-partners the Existential Risk Observatory, AI Forensics, and Data for Good. The conference operationalised Pause IA's positioning as a French civil-society convening organisation rather than only a street-protest movement, and acted as the chapter's substantive policy contribution alongside the Bastille demonstration two days later.

The sibling actions across Europe, North America, and Africa ran on the same demand set — that the Summit programme centre AI safety; that participating governments collaborate on global treaties on frontier-AI training; that an international enforcement body be established to oversee deployment approvals and large training runs; and that national delegations share best practices for protecting their populations from AI-driven harms — coordinated through PauseAI's Discord channels and the @Protest Team and signed up through Luma, Facebook Events, Eventbrite, and Meetup.

Outcome and the AI Action Summit declaration

The Summit closed on Tuesday 11 February 2025 with the Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet, signed by fifty-eight countries — covering accessibility, ethical development, labour-market impacts, and international cooperation. The two non-signatories of particular note were the United States and the United Kingdom: the US declined to sign without a publicly stated specific reason, and the UK declined on the grounds that the Statement's governance provisions were insufficient.

The Summit also launched the Current AI foundation — a public-interest-AI initiative anchored by a €400 million French endowment to support open-source AI tools — and a Coalition for Sustainable AI bringing together eleven countries and thirty-seven companies on the environmental footprint of AI workloads; alongside the EU's InvestAI initiative (a €200 billion mobilisation commitment) and the EU AI Champions Initiative (a €150-billion-over-five-years pledge), the Summit's substantive outputs ran heavily on capability investment and deployment infrastructure rather than the treaty-architecture and binding-rules direction PauseAI had been pressing for.

PauseAI's public reading of the Summit outputs was that the renamed Summit had confirmed the campaign's prior diagnosis — the international Summit series had drifted, in one cycle, from "Safety" (Bletchley 2023) through "Safety" (Seoul 2024) to "Action" (Paris 2025) — and that the Paris Summit had not produced the binding-treaty steps the campaign held to be necessary. The protest wave's value, on the campaign's own reading, was that it had registered a visible international public objection to the Summit's framing in the host city and across four other continents, that it had brought the campaign's first explicit African-continent presence into the campaign record (Kinshasa, Brazzaville, N'Djaména, Yaoundé), and that it had given Pause IA the French civil-society standing the chapter would draw on through 2025 and into the 2026 French municipal-elections charter.

Significance

The Paris protest wave is the moment at which the PauseAI international protests campaign became a five-continent operation. The 13 May 2024 Seoul-eve day of action had been the campaign's first multi-continent action — fourteen cities across Europe, North America, South America, and Oceania, anchored by an OpenAI San Francisco picket and a London Department-for-Science-Innovation-and-Technology demonstration. The Paris wave extended that footprint into Africa for the first time and consolidated the European footprint substantially — eleven European cities against the May 2024 action's eight — while keeping the campaign's working template intact: a single calendar window, a single shared demand set, named-individual hosts where they are available (Fournes and de Scorraille at Bastille; Joseph Miller's Pause AI UK in London; the chapter's own host network across the other cities), and a host-city civil-society conference (Reprendre le Contrôle) anchoring the wave's substantive policy contribution alongside the street action.

It is also the first action in the campaign in which the rename of the international Summit series itself — "Safety" out, "Action" in — became part of the campaign's stated public complaint, the framing on which the 28 February 2026 King's Cross London march, the August 2025 Google DeepMind open letter, and the broader 2025–2026 protest cycle would build. Within the corpus, this entry is the second-anchor (with the May 2024 Seoul-eve action) for understanding how the campaign's international-day-of-action format scales: where Seoul-eve had been the proof of concept, the Paris wave is the build-out — more cities, more continents, more hosts, the same demand set, and the same theory of action that public visibility outside the venue is the route to binding international rules on frontier-AI training that the policy track alone has not produced.