International AI governance: summits, declarations, safety institutes
International AI governance has evolved through high-level summits, joint declarations, and dedicated safety institutes. The Bletchley Declaration and the Hiroshima AI Process set early principles, while the UN's advisory body and Global Digital Compact propose broader frameworks. Safety institutes like the UK's AISI conduct research to inform policy.
The Bletchley Declaration and the AI Safety Summit
The AI Safety Summit, held at Bletchley Park, brought together international governments, leading AI companies, civil society groups, and research experts to consider risks from AI, especially at the frontier of development, and discuss mitigation through internationally coordinated action, as described in the UK Government's AI Safety Summit page.
The summit produced the Bletchley Declaration, which affirms that AI should be designed, developed, deployed, and used in a manner that is safe, human-centric, trustworthy, and responsible. The declaration recognizes that many AI risks are inherently international and best addressed through cooperation, and it resolves to intensify cooperation on frontier AI risks, including in cybersecurity and biotechnology, as stated in the Bletchley Declaration text.
The declaration also notes the potential for serious, even catastrophic, harm from the most significant capabilities of frontier AI models, and affirms the urgency of deepening understanding of these risks and actions to address them.
The Hiroshima AI Process and G7 cooperation
The Hiroshima AI Process was launched in May following the G7 Hiroshima Summit, with the objective of discussing the opportunities and risks of generative AI. After discussions at an interim minister-level meeting and a multi-stakeholder high-level meeting, the Hiroshima AI Process Comprehensive Policy Framework was agreed at the G7 Digital & Tech Ministers' Meeting in December, as explained on the Japan's Hiroshima AI Process page.
This framework includes guiding principles and a code of conduct aimed at promoting safe, secure, and trustworthy advanced AI systems. The process is expected to gather broader support from governments beyond the G7, including developing and emerging economies, as well as the private sector, academia, and civil society, to build inclusive global governance on AI.
The UN's role: advisory body and Global Digital Compact
The United Nations Secretary-General's High-level Advisory Body on AI released its final report, 'Governing AI for Humanity,' in September. The report proposes a blueprint for addressing AI-related risks and sharing its transformative potential, including urging the UN to lay the foundations of a globally inclusive and distributed architecture for AI governance, as detailed on the UN AI Advisory Body page.
The report drew on consultations with more than 2,000 participants worldwide, including 18 deep-dive discussions and over 50 consultation sessions. It proposes seven recommendations to address gaps in current AI governance arrangements and calls on all governments and stakeholders to work together in governing AI.
The Global Digital Compact, negotiated by 193 Member States, is a comprehensive global framework for digital cooperation and governance of AI. It commits governments to upholding international law and human rights online and includes commitments to establish an international scientific panel on AI and a global AI policy dialogue, as stated on the UN Global Digital Compact page.
Safety institutes and international reports
The AI Security Institute (AISI) is the first state-backed organisation dedicated to advancing AI safety. It conducts research and builds infrastructure to understand the capabilities and impacts of advanced AI, and works with the research community, AI developers, and other governments to shape global policymaking, as described on the AISI website.
AISI's goal is to give governments a scientific basis for understanding advanced AI risks, including threats to national security and public safety, and to develop and test mitigations such as safeguards and alignment techniques.
The International AI Safety Report is the first comprehensive global review of scientific knowledge about the capabilities and risks of general-purpose AI. An international Expert Advisory Panel, with members chosen by more than 30 countries and intergovernmental organisations, guided the work, and over 100 independent experts, led by Yoshua Bengio, wrote the report. It provides decision-makers with a shared global picture of AI's risks and impacts, as explained on the International AI Safety Report site.
| Initiative | Type | Primary focus | Adopted |
|---|---|---|---|
| Bletchley Declaration | Declaration | Frontier AI safety risks and international cooperation | At the AI Safety Summit |
| Hiroshima AI Process | Framework | Guiding principles and code of conduct for advanced AI | At the G7 Digital & Tech Ministers' Meeting |
| UN AI Advisory Body | Report | Global governance architecture and recommendations | September |
| Global Digital Compact | Compact | Digital cooperation, AI governance, and human rights | Negotiated by 193 Member States |
| AI Security Institute (AISI) | Safety institute | Technical research and risk evaluation | Ongoing |
| International AI Safety Report | Report | Scientific review of AI capabilities and risks | Annual |
Frequently asked questions
What is the Bletchley Declaration?
The Bletchley Declaration is an agreement by countries attending the AI Safety Summit, affirming that AI should be designed and used in a safe, human-centric, trustworthy, and responsible manner, and recognizing that many AI risks are international in nature.
What is the Hiroshima AI Process?
The Hiroshima AI Process is a G7 initiative launched in May following the G7 Hiroshima Summit to discuss opportunities and risks of generative AI. It produced a comprehensive policy framework with guiding principles and a code of conduct for advanced AI systems.
What is the UN's role in AI governance?
The UN has established a High-level Advisory Body on AI, which released a final report proposing a global governance architecture. The Global Digital Compact, negotiated by 193 Member States, includes commitments to establish an international scientific panel on AI and a global AI policy dialogue.
What does the AI Security Institute do?
The AI Security Institute is a state-backed organization that conducts research and builds infrastructure to understand advanced AI capabilities and impacts, evaluates risks to national security and public safety, and works with developers and other governments to shape policy.
What is the International AI Safety Report?
The International AI Safety Report is a comprehensive review of scientific research on the capabilities and risks of general-purpose AI systems, written by over 100 independent experts and overseen by an international panel nominated by over 30 countries.