Singapore Proposes First Global Standard for Generative AI Testing

Singapore has proposed a new international standard for testing generative artificial intelligence systems. The initiative was announced in a press release by the Infocomm Media Development Authority (IMDA) and Enterprise Singapore (EnterpriseSG).

The proposed standard, ISO/IEC 42119-8, introduces a consistent framework for benchmarking and red teaming methodologies used to evaluate generative AI systems. It aims to improve reproducibility and comparability of test results, providing a foundation for trustworthy AI testing. The proposal will be discussed at the 17th ISO/IEC JTC 1/SC 42 plenary meeting held in Singapore from April 20 to 24, which brings together representatives from over 35 national bodies and more than 250 experts.

The new standard builds on Singapore’s earlier efforts in AI assurance, including the AI Verify Toolkit, the Starter Kit for testing large language model applications, and the Global AI Assurance Sandbox. These initiatives support the development of reliable and safe AI testing methods.

Alongside the plenary meeting, IMDA and EnterpriseSG are organizing several training and policy workshops. These include a foundational training program in collaboration with the American National Standards Institute, an AI standards and policy workshop for national bodies, and the AI Assurance Exchange hosted by the AI Verify Foundation.

Singapore stated that the effort aligns with its broader commitment to building a trusted AI ecosystem through initiatives such as the AI Safety Institute and participation in the ASEAN Working Group on AI Governance.