G7 Hiroshima Code
The bridge between Western frontier labs and governments before binding law arrived; its reporting framework created the first comparable public disclosures of frontier-safety practices across companies.
What it is
The G7's voluntary code of conduct for organizations developing advanced AI: risk assessment, red-teaming, incident reporting, provenance, and safety research investment — with an OECD transparency reporting framework operational since 2025.
OECD-run reporting framework for the Code launched February 2025.
Who is affected
The bridge between Western frontier labs and governments before binding law arrived; its reporting framework created the first comparable public disclosures of frontier-safety practices across companies.
Key obligations
- Voluntary: pre-deployment risk assessment and mitigation, transparency reports, provenance mechanisms
Recent signals
US presses a light-touch AI line at the G20 as the EU proceeds with enforcement
At the G20 digital track the United States argued against binding AI rules while the European Commission continued applying the AI Act — the clearest statement yet that the transatlantic regulatory gap is now policy, not lag.
UN Global Dialogue on AI Governance launches in Geneva
The first universal-membership AI governance forum convenes, alongside the Independent International Scientific Panel established by the August 2025 GA resolution.
UN establishes Independent Scientific Panel and Global Dialogue on AI
The General Assembly creates the first universal AI governance mechanisms, implementing the Global Digital Compact.
Paris AI Action Summit pivots from safety to action
60+ states sign an inclusive-AI declaration; the US and UK decline, and the summit's framing marks the global deregulatory turn.
Related instruments
- INTLOECD AI PrinciplesThe first intergovernmental AI standard (2019), updated in 2024 to address general-purpose and generative AI, misinformation and s…
- INTLUN AI GovernanceFollowing its 2024 AI resolutions and the Global Digital Compact, the UN General Assembly established an Independent International…
- INTLISO/IEC 42001The first certifiable AI management-system standard (the 'ISO 27001 of AI'): organizations implement risk-based AI governance, con…
- INTLAI Summit SeriesThe leader-level summit track evolved from frontier-safety focus (Bletchley Declaration, Seoul frontier-safety commitments) to act…
- EUEU AI ActThe world's first comprehensive, risk-based AI law. It bans a set of unacceptable practices (social scoring, manipulative AI, most…