Overview
On February 10–11, 2025, over 1,000 participants from 100+ countries gathered at the Grand Palais in Paris for the AI Action Summit — the third in the series following Bletchley Park (2023) and Seoul (2024), co-chaired by French President Emmanuel Macron and Indian Prime Minister Narendra Modi.
The summit produced a joint statement signed by 58 countries. The United States and United Kingdom notably refused to sign — a fracture that revealed how quickly the international AI governance consensus assembled at Bletchley had fragmented under geopolitical pressure.
What Happened
The Statement
The “Statement on Inclusive and Sustainable Artificial Intelligence for People and the Planet” emphasized:
- AI as a tool for development, economic opportunity, and closing the global technology gap
- Sustainable AI infrastructure (energy, environmental footprint)
- Interoperability over fragmentation in AI standards
- Preservation of existing international law and human rights frameworks
France’s Strategic Play
Macron used the summit to position France — and Europe — as a neutral third pole between US and Chinese AI dominance:
- Announced a €400M endowment for a new foundation, “Current AI,” with a target of raising €2.5 billion over five years for open AI research
- Hosted CEOs of OpenAI, Google DeepMind, Anthropic, Meta AI, and Mistral
- Framed European AI as “value-aligned” — neither Silicon Valley’s growth-at-all-costs nor China’s state-controlled model
The US Refusal
The Trump administration, inaugurated just three weeks earlier, declined to sign. The US position: innovation over regulation. The same day DeepSeek R1 was released (January 20), Trump had signed Executive Order 14179 revoking Biden’s AI safety order, which had required safety testing disclosures for large AI models. The Paris statement’s emphasis on “sustainable” and “inclusive” AI was seen in Washington as code for regulatory overreach.
The UK Break
The United Kingdom, which had hosted the original Bletchley summit, also did not sign. The Starmer government chose to align with Washington on deregulatory AI policy rather than Brussels — a notable departure from Bletchley’s more expansive multilateralism.
The International AI Safety Report
Published ahead of the summit, the first International Scientific Report on the Safety of Advanced AI — commissioned by the UK government and authored by researchers from 30 countries — documented the state of AI safety science and identified key open problems. The report was received as authoritative by most participating countries, though the US distanced itself from its more cautious framing.
Dario Amodei’s Critique
Anthropic CEO Dario Amodei published a public statement calling the summit a “missed opportunity” — arguing that while AI governance discussions focused on access, inclusion, and economic opportunity, they systematically underweighted the risk of advanced AI systems posing catastrophic or existential risks. The critique highlighted the widening gap between “frontier AI risk” advocates and “AI for development” advocates.
Significance
Paris marked a geopolitical maturation of AI governance: the field moved from the aspiration of a unified global framework (Bletchley) to the reality of competing national strategies. Three distinct governance philosophies were now visible:
- US model: Deregulation, private sector leadership, national competitiveness
- EU model: Rights-based regulation, precautionary principle, mandatory compliance
- Global South/developing world model: AI access, technology transfer, sovereignty over data and deployment
The Paris summit also demonstrated that AI governance had become inseparable from broader geopolitics: decisions about AI standards were simultaneously decisions about technological sovereignty, economic development, and strategic competition.