On 20 April 2026, the Infocomm Media Development Authority (IMDA) proposed an international standard on the quality assessment of prompt-based text-to-text systems that utilise generative artificial intelligence (AI). The proposal outlines approaches to testing and assessing such systems. The standard focuses on establishing consistent methodologies for benchmarking and red teaming, with a view to enabling more structured, comparable, and reproducible evaluations across jurisdictions. It also sets out approaches to testing safety, robustness, and trustworthiness in generative AI applications. The proposal builds on Singapore’s domestic initiatives, including the AI Verify Toolkit.
Original source