Taiwan — Artificial Intelligence Basic Act (2025): AI Compliance Requirements

Taiwan's Legislative Yuan passed the Artificial Intelligence Basic Act on December 23, 2025; it was promulgated and took effect on January 14, 2026. The Act establishes a national framework for responsible AI development and designates the National Science and Technology Council (NSTC) as the central competent authority for AI policy coordination. The Executive Yuan must establish a National AI Strategy Special Committee — chaired by the Premier and comprising scholars, industry representatives, agency heads and local-government leaders — which meets at least once a year. As a framework statute, the Act articulates core principles rather than immediate detailed compliance obligations — and it contains no penalty provisions of its own; sector-specific enforceable rules and penalties are deferred to the relevant ministries, with the Act mandating that within 24 months (by ~14 January 2028) the government review, establish or amend the relevant laws, regulations and administrative measures to conform to the Act. Two concrete implementation steps have since followed: MODA published its AI Risk Classification Framework on July 7, 2026 (announced at the UN's Global Dialogue on AI Governance in Geneva) — a methodology, not a prescriptive prohibited-systems list, that guides competent authorities toward proportionate responses (legislative prohibition/strict control for high-risk applications; guidance/internal controls for lower-risk ones); and the Ministry of Health and Welfare issued the first real sector-specific guidance, "Guidelines for Medical Institutions Applying Generative AI" (醫療機構應用生成式人工智慧指引, MOHW letter 1151663164), on May 29, 2026 — non-binding administrative guidance (not a Basic-Act-mandated enforceable rule) covering medical-record drafting assistance, clinical decision support, administrative documents, and patient communication across three lifecycle stages (pre-adoption assessment, adoption/integration, post-adoption monitoring), while excluding autonomous "AI Agent" systems from its scope. The seven core principles are: sustainability and well-being, human autonomy, privacy and data governance, cybersecurity and safety, transparency and explainability, fairness and non-discrimination, and accountability. AI applications that harm human life, freedom or property, or that undermine social order, national security or the environment, or that involve bias, discrimination, false advertising, misinformation or fabrication, are prohibited under the Act — though the Act itself specifies no enforcement mechanism for those prohibitions.