Taiwan’s Digital Ministry Unveils Draft AI Risk Classification Framework

Article 16 of the Artificial Intelligence Basic Act requires the Ministry of Digital Affairs to develop an AI risk classification framework aligned with international standards, enabling sectoral authorities to identify, assess and address risks. The framework classifies AI by sources of risk rather than imposing centrally determined risk levels. Individual ministries will set the intensity of oversight for applications in areas such as healthcare and finance, balancing industrial innovation with human rights protections.

When the ministry unveiled the draft on April 30, 2026, it listed three broad categories and 18 subcategories. It formally released version 1.0 on July 7, expanding the framework to 20 subcategories covering flaws in technical design, deployment and operation, human-machine interaction, social structures and environmental impacts. The framework recommends four steps: mapping use scenarios, identifying risks, assessing impacts and addressing risks. The government also plans to complete regulatory adjustments within two years after the Artificial Intelligence Basic Act takes effect.