AI Verify Foundation Publishes ISO Mapping for Responsible AI Governance

Organisations seeking a structured approach to AI governance have a new resource to align their work with international standards. On 21 July 2025, the AI Verify Foundation, a not-for-profit subsidiary of Singapore's Infocomm Media Development Authority (IMDA), released a mapping between its AI Verify testing framework and ISO/IEC 42001:2023, the international management system standard for artificial intelligence.

The AI Verify framework, first launched in 2022 as a joint effort by IMDA and the Personal Data Protection Commission (PDPC), provides a set of open-source tools that help organisations assess and document their AI systems against 11 governance principles. The framework covers both traditional AI and generative AI. An updated version addressing generative AI risks was released on 29 May 2025. The new ISO mapping allows organisations to see how the principles and tests in AI Verify correspond to the requirements of ISO/IEC 42001:2023, potentially simplifying compliance and reporting for those pursuing certification or alignment with the international standard.

This development reflects a broader trend in the field of AI governance: the move from voluntary guidelines to concrete, auditable practices. As more organisations deploy AI in regulated sectors such as healthcare, finance, and critical infrastructure, the need for standardised methods to verify that systems are fair, transparent, and accountable has grown sharply. The AI Verify framework, by offering a common set of tests and documentation templates, aims to reduce the burden on individual companies while raising the overall trustworthiness of AI applications.

How the Framework Works

AI Verify is built around 11 governance principles that cover areas such as transparency, explainability, robustness, fairness, accountability, and human oversight. Each principle is accompanied by a set of testable criteria and suggested methods for evaluation. For example, to assess whether a model is biased against a certain demographic group, the framework provides statistical tests and guidance on data collection. To test robustness, it includes adversarial attack simulations and stress-testing procedures.

The tools are open-sourced on GitHub, meaning any organisation can download, modify, and run them locally. This open approach encourages community contributions and allows companies to integrate AI Verify into their existing software development pipelines. The framework is designed to be model-agnostic and can be applied to both traditional machine learning models and large language models used in generative AI.

The mapping to ISO/IEC 42001:2023 is particularly significant because ISO/IEC 42001 is the first international standard to specify requirements for an AI management system. It provides a framework for organisations to establish, implement, maintain, and continually improve an AI management system. By showing how the tests and documentation in AI Verify correspond to the clauses of ISO/IEC 42001, the mapping helps organisations that are already using AI Verify to demonstrate compliance with the standard, and vice versa. This could reduce duplication of effort and speed up the certification process for companies pursuing both.

Why Governance Standards Matter Now

The rapid adoption of generative AI has amplified existing concerns about bias, misinformation, and safety. High-profile incidents where AI systems produced harmful outputs have prompted regulators worldwide to consider new laws and enforcement actions. In this environment, having a tested, transparent, and internationally recognised framework for responsible AI training and deployment is becoming a business imperative rather than a nice-to-have.

For many organisations, the challenge is not a lack of willingness to adopt responsible AI practices but the difficulty of knowing where to start. The landscape of guidelines, principles, and regulations is fragmented. Different jurisdictions have different rules, and non-binding principles from industry groups often overlap or conflict. A framework like AI Verify offers a practical, step-by-step method that an organisation can implement regardless of its size or sector. The open-source nature means that even small companies with limited budgets can access sophisticated testing tools.

The mapping to ISO/IEC 42001 adds another layer of utility. Companies that are already certified to other ISO management standards, such as ISO 9001 for quality or ISO 27001 for information security, will find familiar structure in ISO/IEC 42001. The mapping document explains how each clause of the standard maps to specific tests or documentation in AI Verify, making it easier for internal audit teams and external certifiers to verify compliance.

Workforce Learning and Capacity Building

Effective AI governance is not just about technology; it depends heavily on the skills and knowledge of the people who build, deploy, and oversee AI systems. This is where the concept of responsible AI training becomes central. Organisations need to educate their engineers, product managers, legal teams, and executives on what responsible AI means in practice, how to use testing frameworks like AI Verify, and how to interpret the results.

Workforce learning resources, such as those provided on the site that covers AI Verify, play a key role in bridging the gap between high-level principles and day-to-day work. Training programmes can cover topics such as bias detection, explainability techniques, data governance, and incident response planning. By equipping their teams with these skills, companies can embed responsible AI practices into their development lifecycle from the start, rather than treating them as an afterthought or a compliance checkbox.

The AI Verify Foundation has indicated that it will continue to update the framework as new risks and techniques emerge. The mapping to ISO/IEC 42001 is one of several steps intended to keep the framework relevant and useful. Organisations that invest in responsible AI training now will be better positioned to adapt to future regulatory changes and to build public trust in their AI products.

Looking Ahead

The publication of the ISO mapping is a practical milestone for the AI governance community. It provides a concrete bridge between a widely used open-source testing framework and an international management system standard. For organisations that are already using AI Verify, the mapping offers a clear path to ISO/IEC 42001 certification. For those considering adoption, it adds a strong incentive by demonstrating that the framework is aligned with global best practices.

As AI continues to evolve, the tools and standards used to govern it will need to evolve as well. The AI Verify Foundation's approach of open-sourcing the framework and mapping it to international standards is a model that other governance initiatives may follow. It lowers the barrier to entry for organisations of all sizes and helps create a level playing field where responsible AI practices are the norm, not the exception.