Introduction to Taiwan Artificial Intelligence Basic Act

Introduction to Taiwan
Artificial Intelligence Basic Act

2026/05/15

I.Foreword

Taiwan’s Executive Yuan approved the draft Artificial Intelligence Basic Act (hereinafter, the “Act”) on August 28, 2025, and the Act was passed the Legislative Yuan on December 23. It subsequently was promulgated and entered into force on January 14, 2026. The enactment of the Act represents a significant milestone in Taiwan’s effort to establish a comprehensive governance framework for artificial intelligence.

At its core, the Act pursues a dual objective: to promote technological innovation and industrial development, while simultaneously safeguarding fundamental rights and public interests. It reflects an emerging global consensus that AI governance must strike a careful balance between enabling innovation and mitigating potential risks. To this end, unlike the EU Artificial Intelligence Act, the Act adopts a risk-tiered and principle-based approach, seeking to ensure that AI development and deployment remain aligned with human rights values.

Importantly, rather than imposing immediate and direct regulatory obligations on private entities, the Act focuses on creating an enabling ecosystem through policy guidance, institutional coordination, and incentive mechanisms for research and development. In this sense, the Act serves as a foundational framework law, setting out overarching principles and policy directions, while leaving detailed regulatory requirements to be developed in subsequent legislation and sector-specific rules.

II.The Core Principles

Article 4 of the Act sets out seven fundamental principles governing the research, development, and application of artificial intelligence: human autonomy, sustainability, privacy protection and data governance, safety, transparency and explainability, fairness, and accountability. These principles collectively establish the normative baseline for AI governance in Taiwan and are intended to guide both public policy and future regulatory developments.

From a structural perspective, the Act primarily articulates affirmative (active) and restrictive (passive) expectation on the government authorities.

On the affirmative side, under Article 5 section 2, the government is required to promote AI development in a manner consistent with public welfare. This includes adhering to the principle of prioritizing the best interests of children and adolescents, particularly where AI systems may affect vulnerable groups. Where AI products or systems are designated as high-risk applications by the relevant central competent authority, appropriate precautionary measures and warning mechanisms must be implemented in consultation with the Ministry of Digital Affairs (MODA). In addition, under Article 10, the government is tasked with providing guidance and incentives to foster interdisciplinary innovation across AI-related industries.

The Act further provides direction that the government should actively support the research, development, deployment, and infrastructure of AI technologies under Article 10 subsection 1. This encompasses the formulation of reasonable use policies during the development, training, testing, and impact assessment of emerging AI systems, as well as the provision of financial support, subsidies, and other incentive measures, as stipulated through Article 11 subsection 1.

Another key obligation concerns data governance. Under Article 13, the Act requires the establishment of mechanisms for data openness, sharing, and reuse in order to enhance data accessibility and facilitate AI innovation. Concurrently, the government must regularly review and adjust existing legal frameworks to ensure consistency with evolving technological and societal needs. Article 15 also recognizes the potential labor market impact of AI adoption, mandating the provision of tailored employment counseling and assistance to individuals displaced by automation.

On the restrictive side, under Article 5 section 1, the government is expressly required to avoid and prevent adopting or deploying AI systems that may infringe upon fundamental rights, including life, bodily integrity, liberty, and property. Similarly, AI systems that may endanger social order, national security, or the ecological environment, or that involve bias, discrimination, misinformation, or other unlawful conduct, must be avoided.

In sum, the Act primarily functions as a governance framework directed at public authorities. While it introduces the concept of risk categorization, its practical implementation will depend on the development of detailed standards and enforcement mechanisms. At this stage, the Act establishes the foundation for a risk-managed approach to AI, rather than a fully operational regulatory regime.

III.Relevant Issues Pertaining AI Regulations

The Act does not adopt a sector-specific regulatory model. Instead, it operates as a horizontal, principle-based framework that applies across industries. Consequently, the responsibility for implementation is decentralized: competent authorities in each sector are expected to review, amend, or repeal existing laws and regulations, issue interpretative guidance, and formulate supplementary rules tailored to their respective domains in accordance with Article 18 subsection 1.

Notably, the government has opted not to establish a single, centralized AI regulator. Rather, ministries and agencies retain regulatory authority within their respective jurisdictions, leading to a distributed governance structure. While this approach allows for flexibility and sectoral expertise, it may also give rise to challenges in terms of regulatory consistency and coordination.

Prior to the enactment of the Act, the Executive Yuan had already approved the “Taiwan Artificial Intelligence Action Plan 2.0” (2023–2026), which outlines national policy priorities for AI development.[1] In addition, guidelines on the use of generative AI have been issued for government agencies, signaling an incremental approach to governance through soft law instruments.[2]

From an institutional perspective, different authorities are responsible for overseeing AI-related matters within their respective sectors. For example, AI applications in healthcare fall under the supervision of the Ministry of Health and Welfare, while those in transportation are regulated by the Ministry of Transportation and Communications. Under Article 2 of the Act, the National Science and Technology Council serves as the central competent authority, whereas special municipalities and county, special municipality, or city governments act as local competent authorities. Sector-specific matters, however, remain within the purview of their respective regulators.

To enhance inter-agency coordination, Article 6 requires the Executive Yuan to establish a National AI Strategic Committee. This body is tasked with coordinating national AI policies, overseeing implementation, and formulating the National AI Development Guidelines.

With respect to personal data protection, the Act adopts a sector-based approach. Competent authorities are required to ensure that the collection, processing, and use of personal data in AI development and application are limited to what is necessary and proportionate under Article 14.

IV.Conclusion

The AI Basic Act represents a foundational step in Taiwan’s effort to construct a coherent and forward-looking legal framework for artificial intelligence. By adopting a principle-based and government-led approach, the Act prioritizes policy direction, institutional coordination, and the promotion of innovation over immediate regulatory intervention in the private sector.

At a normative level, the Act underscores Taiwan’s commitment to ensuring that technological advancement remains aligned with fundamental rights and societal values. The articulation of core principles—such as transparency, fairness, and accountability—provides a guiding framework for future legislative and regulatory developments. At the same time, the imposition of both affirmative and restrictive obligations on the government establishes a baseline for responsible public sector adoption of AI.

However, the practical effectiveness of the Act will depend on the timely development of implementing regulations, as well as the ability of different authorities to coordinate across sectors. Given the decentralized regulatory structure, ensuring consistency and preventing fragmentation will be key challenges moving forward.

Looking ahead, further legislative and regulatory initiatives are expected as Taiwan continues to refine its AI governance framework. The long-term success of the Act will ultimately hinge on whether it can meaningfully achieve its intended balance between fostering innovation and protecting fundamental rights in an increasingly AI-driven society.

References

[1] 〈科技部人工智慧科研發展指引108年9月版〉，國家科學及技術委員會，https://digi.nstc.gov.tw/File/7C71629D702E2D89 (The source is only shown in Mandarin.) （最後瀏覽日：2026/5/15）

[2] 〈行政院及所屬機關（構）使用生成式AI參考指引〉，國家科學及技術委員會， https://www.nstc.gov.tw/folksonomy/list/c79bf57b-dc94-4aff-8d14-3262b5559cfc?l=ch (The source is only shown in Mandarin.) （最後瀏覽日：2026/5/15）