Newsletter
Draft Artificial Intelligence Basic Act Comparison Between the Executive Yuan Version and the National Science and Technology Council Version
On August 28, 2025, Taiwan’s Executive Yuan (the cabinet) approved the Draft Artificial Intelligence Basic Act, developed by the Ministry of Digital Affairs ("MODA") following the initial draft by the National Science and Technology Council ("NSTC"). The Executive Yuan subsequently submitted the draft Act to the Legislative Yuan (the congress) for deliberation. Prior to this, on July 15, 2024, the NSTC released its draft of the Act for public comments.
The two versions represent the beginning of the executive branch's version of the Draft Artificial Intelligence Basic Act and its formal submission to the congress for review. The differences in language and policy orientation between these versions provide valuable insight into the legislative evolution of artificial intelligence ("AI") regulation in Taiwan.
I. NSTC Version
The NSTC version focuses on "promoting innovation while balancing human rights and risks," and explicitly outlines seven fundamental principles: sustainable development; human autonomy; privacy protection and data governance; cybersecurity and safety; transparency and explainability; fairness and non-discrimination; and accountability. Furthermore, it sets forth four strategic priorities: foster innovation, collaboration and talent cultivation; risk management and responsible application; protection of rights and data utilization; and regulatory adaptation and operational review.
II. Executive Yuan Version
The version approved by the Executive Yuan retains the foundational principles of the version proposed by NSTC. It sets forth, article by article, the legislative objectives of the Act, definition of AI, fundamental principles for the development and use of AI, and the specific measures that the government should endorse (AI research and application, AI infrastructure; create an environment for AI innovation and experiments; public-private partnerships; international cooperation; cultivate talent; tools and methodologies for evaluation and validation of AI technologies; protect workers' rights; safeguard personal data; preserve multicultural values; ensure intellectual property rights). The draft Act also establishes obligations for government for its use of AI.
In addition, the Executive Yuan version emphasizes “government accountability and legal mechanisms,” stipulating that: (1) when formulating risk-based and tiered regulations, the government shall, in accordance with the law, restrict or prohibit the use of AI in circumstances where such use would constitute a violation of law, (2) the government may, through regulations or guidelines, establish standards, verification, traceability, or accountability mechanisms to enhance the verifiability and human controllability of AI-driven decision-making, and (3) with respect to AI high-risk use cases, the government shall clearly delineate the attribution of responsibility and liability, and establish mechanisms for remedies, compensation, or insurance.
III. Differences Between the Two Versions
First, the NSTC version provides cross-ministerial guidance based on seven fundamental principles and four strategic priorities. The Executive Yuan version further incorporates detailed provisions regarding government responsibilities and specific regulatory mechanisms (such as risk classification, verification, and remedy mechanisms), making it easier to implement future administrative regulations or ministerial rules.
Furthermore, with respect to potential liability and related remedies arising from AI applications, the NSTC version provides general principles regarding accountability and responsible use. The Executive Yuan version advances further by requiring clear attribution of responsibility and liability in relation to high-risk AI, along with the establishment of mechanisms for remedy, compensation, or insurance. These provisions are of substantive significance to individuals and entities affected by AI-related decisions.
The NSTC version’s principle-based framework served as a foundation for public discussions. The Executive Yuan version is grounded in the dual focus on “encouraging innovation” and “balancing interests”, aiming to guide relevant administrative agencies in formulating more institutionalized regulations within their respective domains to address issues arising from the use of AI. This approach seeks to anchor the country’s key developmental priorities, strike a balance between promoting industrial development and risk management, and thereby enhance national competitiveness.
The Executive Yuan’s version of the draft Artificial Intelligence Basic Act is currently under review by the Legislative Yuan. Our firm will continue to monitor the draft status. Should you have any questions regarding the draft Act, please do not hesitate to contact us.