Achieving AI’s global benefits and managing its risks requires broad international cooperation, yet current global debates concentrate disproportionately on the few nations building state-of-the-art AI systems. Singapore shows that smaller, resource-constrained states can influence emerging AI safety norms.
The State of AI Safety in Singapore 2026 report is the second edition of the inaugural report published in 2025. Since 2025, Singapore’s AI safety ecosystem has demonstrated a growing emphasis on implementation-focused governance, with the release of governance frameworks such as the Model AI Governance Framework for Agentic AI. It has also codified testing methodologies and is working to translate them into international and regional standards. This report, current to early May 2026, surveys Singapore’s AI safety ecosystem across four domains: its domestic approach to AI governance, its international approach, its homegrown AI ecosystem and industry, and its technical AI safety research.
We hope this report provides policymakers and industry leaders with valuable insights into Singapore’s pragmatic approach to AI governance and its potential as a regional hub for AI safety testing. AI researchers may also find it useful in identifying opportunities for collaboration with Singapore-based institutions in advancing AI safety science.
This report is intended as a starting point for deeper investigation into specific areas of interest. Given the broad scope, some initiatives are summarized rather than described in full. Readers should treat the findings as a snapshot and consult primary sources for the most up-to-date information.