ISO/IEC 42119-8: Singapore Proposes World's First International GenAI Testing Standard

Singapore has put forward ISO/IEC 42119-8, the world's first international standard for standardizing the testing methodology of Generative AI systems, aimed at strengthening the foundation for trustworthy AI testing. The proposal was announced by the Infocomm Media Development Authority (IMDA) and Enterprise Singapore (EnterpriseSG) on April 20, 2026.

Whats Is ISO/IEC 42119-8? Understanding the GenAI Testing Standard

The proposed ISO/IEC 42119-8 standard focuses on benchmarking and red-teaming methodologies for GenAI systems, aiming to standardize how organizations test AI models for trustworthiness, safety, and reliability.

More specifically, it establishes an important framework for AI testing that enhances the reproducibility and comparability of results, driving assurance and overall trust in AI systems, and enabling safer, more reliable adoption by AI deployers and users.

Key technical pillars of the standard include:

• Benchmarking methodologies — standardized procedures for evaluating GenAI model performance across consistent metrics

• Red teaming methodologies — adversarial testing approaches designed to probe AI systems for weaknesses, biases, and failure modes

• Reproducibility frameworks — ensuring that test results can be replicated across organizations and jurisdictions

  
  

Why Singapore? The Road to ISO/IEC 42119-8

Singapore's proposal did not emerge in a vacuum. ISO/IEC 42119-8 builds on IMDA's past work in developing domestic testing frameworks, including the AI Verify Toolkit, the Starter Kit for Testing of LLM-Based Applications for Safety and Reliability, and the Global AI Assurance Sandbox.

The Sandbox, run under the AI Verify Foundation, is already testing AI systems against real-world problem statements, generating findings that IMDA views as "pre-standardization material" that can feed into broader international efforts.

As IMDA Chief Executive Ng Cher Pong stated at the plenary opening: "Standards are the quiet infrastructure that enables interoperability, consistency and trust at scale." He also warned that "standards setting cannot move at a glacial pace," cautioning that overly slow processes risk irrelevance, a pointed message as AI evolves from generative to agentic systems.

The 17th ISO/IEC JTC 1/SC 42 Plenary: A Historic Meeting

The standard is being discussed at the 17th ISO/IEC JTC 1/SC 42 plenary meeting, held in Singapore from April 20–24, 2026, co-organized by IMDA and EnterpriseSG and hosted in the ASEAN region for the first time.

The meeting 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.

SC 42 is the subcommittee for AI within the ISO/IEC Joint Technical Committee (JTC 1), responsible for developing international standards on AI covering foundational standards, trustworthiness, use cases, data, and testing.

Singapore's Broader AI Standards Commitment

ISO/IEC 42119-8 is one piece of a much larger strategy. It is part of Singapore's broader commitment to international AI standards, as seen in the national adoption and accreditation program of ISO/IEC 42001 led by EnterpriseSG, and the contribution of real-world use cases to support ISO/IEC TR 24030's documentation of AI applications in practice.

This push toward standardization also comes alongside broader efforts to strengthen AI governance frameworks in Singapore. Earlier in 2026, IMDA unveiled a new Model AI Governance Framework for Agentic AI, designed to guide the safe deployment of autonomous AI systems that can reason and take actions on behalf of users.

Internationally, Singapore is committed to developing a trusted AI ecosystem through its AI Safety Institute (AISI), leadership in the ASEAN Working Group on AI Governance (WG-AI), and its role as lead ASEAN member for the AI Priority Area under the ASEAN Digital Trade Standards and Conformance Working Group (DTSCWG).

Capacity-building workshops were also hosted alongside the plenary, including:

• A Broad-Based Foundational Training Workshop with the American National Standards Institute (ANSI) for ASEAN member states

• An AI Standards and Policy Workshop targeted at national standards bodies and AI policymakers from 15 countries

• An AI Assurance Exchange organized by Singapore's AI Verify Foundation, covering AI standards opportunities, gaps, and regional implementation

What This Means for Manufacturers and AI Developers

For companies building, deploying, or integrating generative AI systems, ISO/IEC 42119-8 signals a significant shift in the regulatory landscape:

• Testing will become standardized globally. Organizations that currently use proprietary or ad hoc AI testing methods should begin aligning with benchmarking and red teaming approaches consistent with the proposed standard.

• Reproducibility becomes a compliance requirement. AI systems will need to demonstrate that testing outcomes are consistent and comparable across different environments and jurisdictions — not just internally validated.

• Market access implications. As more countries align with ISO/IEC standards, manufacturers seeking to sell AI-powered products in regulated markets (healthcare, finance, critical infrastructure) may be required to demonstrate conformance to ISO/IEC 42119-8 testing protocols.

• Red teaming goes mainstream. What was once a niche security practice will become a formal, standardized requirement for AI system validation.

• Documentation and traceability. Organizations will need audit-ready records of their testing methodologies to demonstrate alignment with the standard.

  
  

Certification Impact Summary

Aspect	Current State	Post-ISO/IEC 42119-8
GenAI Testing Method	Ad hoc / proprietary	Standardized benchmarking + red teaming
Cross-border comparability	Limited	Enhanced reproducibility across jurisdictions
Certification basis	No global standard exists	ISO/IEC 42119-8 framework
Existing frameworks	ISO/IEC 42001 (AI Management Systems)	42001 + 42119-8 (testing layer added)
Scope	AI management systems	Specifically GenAI testing methodology
Trust & assurance	Organization-defined	Globally recognized baseline

Organizations already certified under ISO/IEC 42001 (AI Management Systems) will find 42119-8 a complementary layer — extending governance frameworks into the testing and validation domain specifically for generative AI.

Timeline + Required Actions

Date	Milestone
April 20, 2026	IMDA/EnterpriseSG announce ISO/IEC 42119-8 proposal
April 20–24, 2026	Proposal discussed at 17th ISO/IEC JTC 1/SC 42 plenary, Singapore
TBD	ISO/IEC working group review and draft development
TBD	Public comment / ballot period
TBD	Final publication of ISO/IEC 42119-8

Recommended Actions for Organizations

1. Monitor the ISO/IEC 42119-8 lifecycle page at iso.org for draft availability and comment periods.

2. Audit current GenAI testing practices against benchmarking and red teaming best practices — gaps identified now are easier to close before the standard is finalized.

3. Engage your national standards body (e.g., ANSI in the US, BSI in the UK, SAC in Singapore) to participate in the comment and review process.

4. Review ISO/IEC 42001 alignment — if not yet certified, consider beginning the process now to ensure the management system foundation is in place before 42119-8 testing requirements take effect.

5. Brief internal AI/ML teams on the red teaming and benchmarking requirements likely to emerge from the standard.