Despite growing geopolitical tensions, an unexpected convergence is emerging on the governance of extreme AI risks that are universal, irreversible, unpredictable and potentially catastrophic. The Strategic Foresight Group report, The Essential Convergence: Global Compact on Extreme AI Risks, draws on expert consultations across several regions and identifies areas where countries often seen as strategic competitors are beginning to align around common safeguards for advanced AI.
First, it shifts the core question from what should be done for AI risks governance to what states are already doing for extreme AI risk mitigation that can contribute to peace and security. It shows that countries as diverse as United States, China, the European Union, India, Brazil, South Africa, South Korea and the UAE, are beginning to converge around common safeguards which would include metrics of frontier models, pre-deployment evaluation, rigorous safety testing, frontier weight protection, and incident reporting.
Second, rather than advocating a centralised global regulatory regime, the report proposes a compact built around minimum safeguards to prevent two ultimate risks: AI systems that enable weapons of mass destruction and AI systems that could lead to irreversible loss of human control.
Third, the report challenges the widespread assumption that only a handful of technology powers can shape the future of AI. It demonstrates that demand-side countries controlling markets, ingredients, labour, energy, and procurement systems possess significant leverage and can play a decisive role in establishing global safeguards, even if they do not develop frontier AI systems themselves.
The launch will bring together leaders from the humanitarian, diplomatic, parliamentary, scientific and security communities, to examine whether the world can learn from successful international initiatives of the last century—such as the Red Cross, landmine ban, arms control agreements—to propose a practical framework for international cooperation on extreme AI risks before a major crisis compels action.
Speakers
- Jagan Chapagain, Secretary General, IFRC
- Thomas Greminger, Executive Director, GCSP
- Jody Williams, Nobel Peace Laureate
- Sundeep Waslekar, President, Strategic Foresight Group
- Melissa Parke, Executive Director, ICAN
- Chinmay Pandya, Pro Vice Chancellor of Dev Sanskriti Vishwavidyalaya University, India
- Anda Filip, Secretary General, Inter-Parliamentary Union
- Lavina Ramkissoon, Co-Chair, AU-ASRIC Council (AI), African Union
- Derrick Swartz, Science Expert, Ministry and Department of Science, Technology and Innovation, South Africa
- Martin Müller, Executive Director, GESDA
- Kwan Yee Ng, Head of International AI Governance, Concordia AI, China
Moderator
- Anne-Marie Buzatu, Executive Director, ICT4Peace Foundation
Anchor
- Christina Schori Liang, GCSP
