Paris hosts the third AI summit and the regulatory consensus that emerged at Bletchley quietly fractures

On 10 and 11 February 2025, France hosted the third in the international AI Summit series, after Bletchley in November 2023 and Seoul in May 2024. The Paris AI Action Summit gathered representatives from sixty governments, including direct attendance from US Vice President JD Vance, French President Emmanuel Macron and Indian Prime Minister Narendra Modi. Macron used the opening day to announce a 109-billion-euro sovereign-AI investment package, drawing on private and pension-fund capital, and to position France as the European AI infrastructure hub.

The substantive output was the Statement on Inclusive and Sustainable Artificial Intelligence, signed by sixty-one parties including the European Union, China, India and most of the Global South delegations. As Politico Europe and the Financial Times reported the same evening, the statement notably was not signed by the United States or the United Kingdom. The Vance speech the same morning had set the tone: the United States, the speech argued, would prioritise AI competitiveness over what the speech called "foreign-jurisdiction safety theatre".

Three subplots from the summit

First, the French sovereign-AI announcement. The 109-billion-euro figure was assembled mostly from existing private-capital commitments and pension-fund allocations, with a smaller direct state component. As reported by Le Monde and the FT, the headline number was substantially aspirational, but the underlying datacentre-and-energy commitments at Cadarache and elsewhere were concrete. Second, the US-UK refusal to sign was a sharper rupture from the Bletchley framework than most observers had anticipated. Third, the China delegation's substantive engagement on the safety chapter, which had been the diplomatic surprise at Bletchley, persisted.

Sixty signatures, two refusals, and a chair speech that read like a closing argument.

What was actually consequential

The fracture between the Bletchley-style consensus framework and the US-UK position is the lasting outcome. The next eighteen months produced a parallel-track international regulatory environment: the EU and China negotiating bilaterally on standards interoperability, the United States moving forward with bilateral arrangements (notably with Japan, Korea and Australia), and the United Kingdom's softer divergence working its way toward what would become the UK AI Bill of March 2026. The single global frontier-AI conversation of late 2023 had become, by mid-2025, a three-bloc conversation.