MAS FEAT principles for Singapore insurers
MAS published its Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of Artificial Intelligence and Data Analytics in 2018. These principles apply to all MAS-regulated financial institutions using AI in consequential decisions — including insurers. The principles cover: Fairness (AI does not produce discriminatory outcomes); Ethics (AI use aligns with ethical values and does not harm customers); Accountability (clear responsibility for AI systems and their outcomes); and Transparency (AI decisions can be explained to customers and regulators).
For insurance, the most significant FEAT applications are: underwriting AI (fairness in risk assessment, avoiding discriminatory pricing); claims AI (fair and consistent claims assessment, explainability of claims decisions to policyholders); fraud detection AI (minimising false positives that unfairly treat legitimate claimants as fraudsters); and distribution AI (suitability of product recommendations).
Veritas framework for insurance AI
The Veritas Consortium — a MAS-supported industry initiative — developed assessment methodology for implementing the FEAT principles. Phase 1 covered credit risk scoring and customer marketing; Phase 2 developed frameworks for fraud detection. Singapore insurers using AI in these contexts should assess their AI systems against Veritas methodology, both as a compliance demonstration and as a governance best practice. MAS has indicated that Veritas compliance will be considered positively in supervisory assessments of AI governance maturity.
PDPA obligations for insurance AI
Insurance involves processing substantial volumes of sensitive personal data — health information in life and health underwriting, claims history, financial information, and increasingly telematics and behavioural data. The PDPA's obligations apply to all of this: consent or a legitimate purpose exception for each use of personal data in AI; purpose limitation (data collected for one purpose cannot be used in AI for another without consent); security safeguards; and individual access and correction rights.
Health data used in life insurance underwriting is sensitive personal data under the PDPA and attracts higher obligations. Insurers must have documented lawful basis for processing health data in AI models and must be able to demonstrate that processing is necessary for the purpose claimed. The PDPC has been active in enforcing data protection standards in financial services contexts.
