Singapore’s foreign minister called for a UN framework convention on AI safeguards, urging international cooperation to set safety standards for frontier AI models.

Singapore’s foreign minister has urged the United Nations to adopt a global framework convention that would set safety standards for frontier artificial intelligence models, marking a significant diplomatic push for coordinated AI governance.

Call for a UN‑wide AI Safety Convention

At a recent multilateral forum, Singapore’s foreign minister emphasized that the rapid development of advanced AI systems poses cross‑border risks that no single nation can manage alone. He advocated for a binding UN convention that would establish baseline safeguards, transparency obligations, and mechanisms for international oversight.

Key Elements of the Proposed Framework

  • Mandatory risk assessments for high‑impact AI models before deployment
  • Standardized reporting of AI capabilities and training data to a UN registry
  • Shared protocols for incident response and mitigation across member states
  • Periodic review cycles to adapt standards to emerging technologies

The proposal calls for participation from both AI developers and governments, aiming to balance innovation with public safety. It also suggests the creation of an expert panel within the UN to guide the implementation of technical standards.

International Reactions

Several countries have expressed support for a coordinated approach, citing the need for a level playing field and the prevention of an AI arms race. However, some nations remain cautious, warning that overly prescriptive rules could stifle research and economic growth.

The initiative aligns with broader global efforts, such as the G20 AI principles and the OECD’s AI policy recommendations, but seeks to embed these guidelines within a formal UN treaty structure.

“AI safety is a shared responsibility that transcends borders; a UN convention would provide the collective resolve we need,” the minister said.

The proposal is expected to be tabled at upcoming UN discussions on emerging technologies, where member states will debate the scope and enforceability of such a convention.

The Straits Times coverage of Singapore's AI safety proposal