Singapore: Adopted Second Edition Model AI Governance Framework

On 21 January 2020, the Infocomm Media Development Authority (IMDA) of Singapore, together with the Personal Data Privacy Commission (PDPC) and in collaboration with the World Economic Forum (WEF), adopted the Second Edition Model AI Governance Framework. The Governance Framework updates the First Edition Model AI Governance Framework, adopted in 2019, and covers internal governance structures and measures, determination of AI decision-making models, operations management, and customer relationship management. The second edition provides industry examples in each section to illustrate how organisations have implemented AI governance practices. The updated Framework further clarifies certain definitions, such as the human-over-the-loop approach, which covers human supervision in augmented decision-making. Moreover, the framework clarifies that organisations can consider factors such as the nature and reversibility of harm and operational feasibility in determining the level of human involvement in an organisation’s decision-making process involving AI. Lastly, the updated framework provides guidance on the necessity and relevance of measures, such as including personal and non-personal data when building AI models and introducing robustness, reproducibility, and auditability as criteria.