AI Guide For Boards in Singapore

AI GUIDE FOR BOARDS IN SINGAPORE

Section III: Governance and Oversight

7.3 Integrating AI into Enterprise Risk Management

7.3.1 Recalibrating the risk appetite

AI risk cannot be managed as a standalone technical issue. Boards should ensure that AI-related risks are integrated into the organisation’s existing enterprise risk management framework. Rather than creating an isolated risk category, boards should require management to assess how AI amplies or reshapes existing risk exposures, including: • Operational risk: Systemic model failures, AI dri, data poisoning, third-party dependencies or disruption to critical business processes. • Strategic risk: Poor capital allocation, unscalable AI pilots, over-reliance on opaque systems (such as black-box optimisation models), competitive disruption, or misalignment with corporate values. • Legal and compliance risk: Direct infringement of intellectual property, data privacy violations, and non-compliance with shiing local or cross-border AI regulations. • Reputational risk: AI bias, automated discrimination, hallucinated outputs, misleading communications, and systemic erosion of stakeholder trust. Boards should ensure that AI-related risks are reected within enterprise risk registers, escalation pathways, control frameworks and assurance processes.

Boards should review the organisation’s risk appetite statement to incorporate specic boundary conditions for AI experimentation, deployment and operational use.

is falls into two broad categories: 1. Zero-tolerance thresholds: Clear

boundaries where fully autonomous AI deployment is strictly prohibited without human oversight or specic board approval. For example, fully automated employee terminations, unsupervised high-value nancial transactions, and regulatory lings without human verication. 2. Acceptable technical and governance debt: e extent to which the organisation is willing to tolerate explainability gaps, legacy systems, incomplete controls, or delayed governance remediation in exchange for speed or operational exibility. Boards should ensure that any accepted governance or technical debt is explicitly documented, time-bound, monitored and supported by remediation plans.

Made with FlippingBook Digital Proposal Creator