AI GUIDE FOR BOARDS IN SINGAPORE
Section III: Governance and Oversight
7.1.3 Frontline customer-facing
Board checklist for management • Where do AI systems directly interact with customers or external stakeholders? • Are there disclaimers in place for the interaction? • What guardrails, testing and monitoring controls exist before and a er deployment? • What escalation and fallback mechanisms exist when AI systems fail? • How are AI-speci c vendor and supply chain risks managed contractually and operationally? stakeholders. Rigorous guardrails, pre- deployment testing should be conducted and monitored post-deployed. Escalation and human fallback processes should be de ned, and crisis response protocols for AI-related incidents set out clearly. As competitors and threat actors are also changing their AI use, boards should also require management to increase resilience against such changing competition and misuse
• Opportunities: Customer-facing AI systems may signi cantly improve service responsiveness and customer engagement, with customer service chatbots, AI- generated marketing content, algorithmic pricing, automated service distribution and autonomous supply chain automation. is lowers interaction friction and scales corporate outreach. • Risks: External-facing AI systems typically carry the highest level of legal, reputational, operational and regulatory exposure. Failures and system errors are o en immediately visible to customers, regulators, investors and the public. Key threats include inaccurate or misleading outputs (such as hallucinations or inaccurate responses that misrepresent products or contract terms). AI- speci c cyber threats introduce new and heightened vulnerabilities, while third-party and supply chain risks may expose organisations to unintended risk outcomes. • Board considerations: Boards should require management to identify where AI systems directly interface with external
60
Made with FlippingBook Digital Proposal Creator