Taiwan Plans Four-Tier AI Risk Controls for Government

Taiwan’s Ministry of Digital Affairs is developing a standardized artificial intelligence risk-assessment system for government agencies under the AI Basic Act. The framework would divide AI applications into four risk tiers and apply different governance requirements to each. The initiative matters because it would give public bodies a common structure for identifying risks, assigning accountability and protecting privacy, security and citizens’ rights as government use of AI expands.

The ministry plans to publish an internal-control template for government AI systems in October, allowing agencies to adopt safeguards based on each application’s risk level. It is also considering requiring large government AI services to undergo risk reviews before launch. Once deployed, those systems would face regular rolling assessments so controls can be adjusted as their use, data inputs and external risks change over time.