Hiroshima AI Process
G7 framework for responsible AI development, adopted at the 2023 Hiroshima summit under Japanese presidency.
Last refreshed: 3 June 2026 · Appears in 1 active topic
Why did France lean on the Hiroshima AI Process instead of its own cloud doctrine?
Timeline for Hiroshima AI Process
Retained as the operative AI governance framework in the signed G7 communique
European Tech Sovereignty: France's G7 text drops cloud sovereignty- What is the Hiroshima AI Process and is it legally binding?
- The Hiroshima AI Process is a voluntary G7 framework for responsible AI, adopted in May 2023. Its principles and code of conduct for AI developers are non-binding; countries implement them domestically at their own discretion.Source: Briefing event
- Why did the 2026 G7 digital ministerial use the Hiroshima AI Process framework?
- The Bercy ministerial communique of 29 May 2026 anchored its AI commitments to the Hiroshima Process because the voluntary, non-binding framework was acceptable to all G7 members including the United States, unlike France's own cloud sovereignty doctrine which Washington objected to.Source: Briefing event
- Who are the signatories of the Hiroshima AI Process?
- The seven G7 members (Canada, France, Germany, Italy, Japan, the United Kingdom and the United States) plus the European Union adopted the Hiroshima AI Process in 2023.Source: Briefing event
Background
The Hiroshima AI Process is an international AI-governance framework adopted by the G7 under Japan's 2023 presidency, at the Hiroshima Leaders' Summit in May 2023. It established voluntary international guiding principles for advanced AI systems and a voluntary code of conduct for AI developers, covering transparency, safety testing, cybersecurity risk management and the labelling of AI-generated content. The Process was designed to align G7 members on responsible AI development without imposing binding legal obligations, positioning it as a complement to domestic regulation such as the EU AI Act rather than a replacement. Signatories are the G7 members: Canada, France, Germany, Italy, Japan, the United Kingdom and the United States, plus the European Union.
The G7 Digital and Technology Ministerial Declaration signed at Bercy, Paris on 29 May 2026 under French chairmanship listed the Hiroshima AI Process as one of its four agreed reference frameworks, notably in the absence of any mention of cloud sovereignty, CAIDA or restrictions on US cloud providers . France's choice to anchor the communique to the Hiroshima AI Process rather than its own 'Cloud au Centre' doctrine was read as a signal that Paris had accepted Washington's framing for the ministerial text. The process's voluntary, non-binding character made it politically acceptable to all G7 members precisely because it imposed no procurement preferences or data-residency mandates that US firms could object to under trade frameworks.
