AI GUIDE FOR BOARDS IN SINGAPORE
Section V: Organisational Resilience and Capability
11.2 Data Strategy
As AI adoption scales, boards need to have oversight of how data is sourced, governed, protected, accessed, monitored and used to train or operate AI systems. is includes ensuring that AI systems remain secure, auditable, explainable and legally defensible over time. Board oversight should extend across enterprise data strategy, data quality and integrity, access governance, model lifecycle management, operational monitoring and intellectual property considerations relating to AI-generated outputs.
trustworthiness of AI systems and reduce operational and regulatory risk. Data governance also strengthens organisational resilience against cyber, legal and reputational exposure. Poor-quality or poorly-governed data can lead to inaccurate outputs, biased decisions, regulatory breaches and loss of stakeholder trust.
Boards should recognise that weakness in data governance frequently becomes weakness in AI governance.
Strong governance in these areas helps organisations improve the reliability and
11.3 Security and Accessibility
Data governance requires boards to have oversight of data classi cation, access controls, data residency and data lineage.
AI performance depends on trusted, accessible and well-governed data foundations.
11.3.1 Data classi cation
Boards have to balance between security (protection of sensitive data, customer trust and regulatory compliance) on the one hand, and accessibility (access to data to support scalable AI development) on the other hand.
Organisations should classify data according to sensitivity and business criticality to support appropriate governance and access controls. Data Sensitivity Classi cation: Examples include:
96
Made with FlippingBook Digital Proposal Creator