New ISO proposal from Singapore focuses on benchmarking and red teaming to improve trust in AI systems worldwide.

Singapore has introduced a new international standard aimed at improving how Generative AI systems are tested. The proposal, ISO/IEC 42119-8, was announced on 20 April 2026 and will be discussed at the 17th plenary meeting of ISO/IEC JTC 1/SC 42 in Singapore, held from 20 to 24 April.

The meeting is co-organised by the Infocomm Media Development Authority and Enterprise Singapore. It is the first time the event is hosted in the ASEAN region and brings together more than 250 AI experts and representatives from over 35 countries, including the United States, United Kingdom, China, Japan, Germany, France, and South Korea.

The proposed ISO/IEC 42119-8 standard focuses on benchmarking and red teaming methods for Generative AI systems. It aims to create a common testing framework that improves consistency, reliability, and comparability of AI results. This is expected to help organisations ensure their AI systems are safe, trustworthy, and ready for wider adoption.

Singapore’s proposal builds on earlier work by IMDA, including tools such as AI Verify and testing frameworks for large language model applications. It also supports broader international efforts like ISO/IEC 42001 and ISO/IEC TR 24030, which focus on AI management systems and real-world applications. These initiatives aim to strengthen global confidence in AI technologies.

Alongside the plenary meeting, several workshops and events are being held to build AI standards capabilities. These include a training workshop organised with the American National Standards Institute, as well as an AI Standards and Policy Workshop led by ISO. The AI Verify Foundation is also hosting the AI Assurance Exchange, where global experts will discuss challenges and opportunities in AI standards and governance.

Singapore continues to play a key role in global AI governance through initiatives such as its AI Safety Institute and leadership in ASEAN AI working groups. These efforts aim to ensure that AI systems are developed and used in a safe, reliable, and trustworthy way for businesses and individuals.

Source: Entrprise SG