Taiwan's AI Basic Act: Comprehensive Article-by-Article Analysis

Dr. Ju-Chun Ko Dr. Ju-Chun Ko
閱讀中文版本

Disclaimer: This discussion is based on the “Artificial Intelligence Basic Act - Integrated Version of Proposals by 12 KMT Legislators.” The bill was reviewed article by article at the joint meeting of the Legislative Yuan’s Education and Culture Committee and Transportation Committee on August 4, 2025. The final version passed by the committee shall be subject to the official parliamentary records.

Update: According to the results of the joint review meeting of the Legislative Yuan’s Education and Culture Committee and Transportation Committee on August 4, 2025, the bill’s title and Articles 1 to 11 remain reserved based on previous review results. Articles 13, 16, 17, 18, 19, 20, and chapter titles, along with Articles 33 to 39 proposed by the Taiwan People’s Party caucus, are reserved. The bill draft (passed by committee) has been sent to the plenary session for processing.

Foreword: The Global AI Regulatory Race and Taiwan’s Legislative Urgency

2025 has seen countries accelerating their AI legal frameworks: the EU’s AI Act is coming into effect, Japan passed its AI Promotion Act in late June, the US White House released its AI Action Plan in July, and China announced its Global AI Governance Action Plan in July. The overall trend shows international consensus on balancing innovation promotion and risk mitigation, establishing AI development norms through high-level strategic coordination and principled guidance.

Without enacting comprehensive legislation, Taiwan risks not only constraining its technological development due to unclear legal foundations but also being excluded from international regulatory circles. This session’s Artificial Intelligence Basic Act integrates proposals from almost 20 cross-party legislators, aiming to legally establish government AI development policies while balancing innovation promotion and ethical governance in parallel, demonstrating “Taiwan Speed” in the global AI competition.

Article 1: Legislative Purpose and Objectives

Article Content: Article 1 outlines the legislative purposes, including: building a smart nation, promoting AI R&D and industrial development, constructing safe AI application environments, implementing digital equity, protecting fundamental civil rights, enhancing social welfare, improving quality of life, promoting sustainable national development, preserving national cultural values, enhancing international competitiveness, and ensuring technology applications align with social ethics.

Conceptual Analysis: This article consolidates multiple national and social objectives for AI development, serving as the charter for the entire law. First, it highlights AI’s importance to economic and social development, positioning AI as a core driver for national development and competitiveness enhancement. Similar language appears in Japan’s AI Promotion Act, where the Japanese Diet emphasized “AI-related technologies as fundamental technologies for Japan’s economic and social development.”

This positioning demonstrates legislators’ determination for AI to flourish domestically. Second, the article includes social welfare, sustainable development, digital equity, civil rights, and cultural values, showing that legislation is not technology for technology’s sake, but ensuring AI development serves the welfare of all citizens and Taiwan’s cultural identity.

The inclusion of digital equity and fundamental rights protection reflects legislators’ concerns about potential digital divides and human rights challenges brought by AI, emphasizing that AI development cannot sacrifice vulnerable groups’ rights and democratic values.

Article 2: Competent Authority Designation

Article Content: Article 2 designates the National Science and Technology Council (NSTC) as the central competent authority, with local governments as regional authorities; matters involving other competent authorities shall be handled by the respective authorities.

Conceptual Analysis: This article establishes the primary agency for AI development. Choosing NSTC (an Executive Yuan agency specializing in technological development and R&D) as the central authority embodies a development and innovation-first stance rather than pure regulatory thinking.

This aligns with international mainstream practices: Japan’s AI Promotion Act is also led by its technology policy system, establishing a Prime Minister-led AI Strategy Headquarters after passage; South Korea also established a National AI Committee in its AI Framework Act. In contrast, if Taiwan were led by the Ministry of Digital Affairs (a third-level agency), its limited resources and cross-ministerial coordination capacity might bias toward control while neglecting R&D.

Article 3: Definition of Artificial Intelligence

Article Content: Article 3 defines “artificial intelligence” as “machine-based systems that, operating with varying degrees of autonomy, may exhibit adaptiveness after deployment, generating predictions, content, recommendations, or decisions through sensing or input to achieve specific human-defined purposes.”

Conceptual Analysis: Establishing a clear AI definition helps clarify the law’s scope of application. This definition adopts international standard practices, closely resembling the EU AI Act’s description of AI systems. The EU legislative draft also defines AI systems as machine-based systems capable of generating content, predictions, or decisions, emphasizing characteristics like autonomy and adaptiveness.

This article adopts a broader and technology-neutral definition, aiming to enhance the law’s foresight and inclusiveness: neither limiting to specific technical routes nor becoming outdated due to technological evolution.

Article 4: Basic Principles for Government AI Promotion

Article Content: Article 4 regulates that governments promoting AI R&D and applications must balance social welfare, digital equity, innovation R&D, and national competitiveness while developing good governance and infrastructure, following eight basic principles:

  • Sustainable Development and Welfare: Balancing social equity and environmental sustainability, providing appropriate education and training to reduce digital divides
  • Human Autonomy: Supporting human autonomy, personality rights, and basic rights and cultural values, allowing human oversight
  • Privacy Protection and Data Governance: Properly protecting personal data privacy while promoting open reuse of non-sensitive data
  • Intellectual Property Protection: AI development and applications must comply with intellectual property regulations
  • Cybersecurity and Safety: Establishing cybersecurity measures and threat prevention
  • Transparency and Explainability: AI outputs should undergo appropriate information disclosure or marking
  • Fairness and Non-discrimination: Avoiding algorithmic bias and discrimination risks
  • Accountability: Ensuring relevant entities or developers bear appropriate internal governance and external social responsibilities

Conceptual Analysis: These eight principles establish the ethical foundation for AI good governance. Their content highly aligns with AI ethical principles proposed by advanced countries and international organizations.

Article 5: Digital Equity and Vulnerable Group Protection

Article Content: Article 5 stipulates that people, regardless of gender, age, ability, region, ethnicity, beliefs, political views, or socioeconomic status, should equally enjoy opportunities to access AI; for indigenous peoples, disabled individuals, and other vulnerable groups, their autonomy and special characteristics should be considered, providing special protection and development assistance according to law.

Conceptual Analysis: This article’s core is the digital inclusion concept manifested in the AI era. If AI technology is only mastered by some people, it may widen digital divides and exacerbate social inequality. Legislating equal AI access rights for all demonstrates the goal of technology democratization.

Article 6: Preventing AI Misuse and Minor Protection

Article Content: Article 6 requires the government to prevent AI applications from causing harm to people’s lives, bodies, freedom, property, damaging social order or ecology, and bias discrimination, false advertising, information misleading, or counterfeiting. For minor protection enhancement, it stipulates that AI products or systems recognized as high-risk by competent authorities in consultation with the Ministry of Digital Affairs should clearly mark precautions or warnings.

Conceptual Analysis: This article belongs to risk prevention and ethical bottom-line categories, ensuring AI technology isn’t misused to harm public interests and vulnerable groups. The article directly incorporates these risks into the basic law, declaring the government’s determination to strictly prevent AI from becoming a new tool for illegal crimes or rights infringement.

Article 7: National AI Strategy Special Committee

Article Content: Article 7 requires the Executive Yuan to establish a National AI Strategy Special Committee, convened by the Premier, with members including scholars and experts, AI-related civil organizations and industry representatives, political affairs commissioners, relevant ministry heads or representatives, and mayors of municipalities and counties.

Conceptual Analysis: This article establishes a high-level AI governance coordination mechanism. The Premier’s personal convening means AI strategy rises to the highest national decision-making level. The committee’s diverse membership composition, covering industry, government, academia, and research plus local governments, ensures broad decision-making perspectives.

Article 8: Government-Led Digital Sovereignty and AI Applications

Article Content: Article 8 stipulates that to accelerate building AI that aligns with our country’s characteristics and national AI strategy, ensuring our digital sovereignty and effectively enhancing government efficiency, government agencies should actively promote AI deployment and applications.

Conceptual Analysis: This article emphasizes the importance of government as an AI application pioneer. On one hand, widespread government AI adoption can improve public service efficiency and reduce costs. On the other hand, the deeper intention is defending Taiwan’s digital sovereignty and cultural autonomy.

Article 9: AI Education and Universal Digital Literacy

Article Content: Article 9 requires the government to continuously promote AI education across schools at all levels, industries, society, and government agencies, enhancing citizens’ digital literacy and AI knowledge skills. The central education authority should formulate AI usage and learning guidelines for schools at all levels.

Conceptual Analysis: This article focuses on talent cultivation and digital capacity building. In the AI era, the most important capital is people. If citizens lack AI literacy, they not only cannot utilize new technology well but are easily affected by its negative impacts.

Article 10: Cross-Domain Cooperation and International Integration

Article Content: Article 10 requires the government to implement AI development policies and encourages industry, government, and academia to actively engage in cross-domain cooperation and international exchanges in eight key areas including AI R&D innovation, basic education, professional talent cultivation, industry development, industry-academia cooperation, investment encouragement, infrastructure building, and other AI development policies.

Conceptual Analysis: This article consolidates the government’s eight action directions for AI development, outlining the main territory of national AI strategy. These eight points cover everything from education, talent, R&D to industry, capital, infrastructure, representing a continuation and refinement of the 2018 Taiwan AI Action Plan.

Article 11: Budget Guarantee

Article Content: Article 11 requires the central government to generously allocate budgets within national fiscal capacity and take necessary measures to continuously ensure funding meets AI policy development needs.

Conceptual Analysis: This clause is concise but crucial—without budget, even the best policies remain paper tigers. This article uses legal force to require the government to adequately allocate AI budgets, significantly elevating AI programs’ status in annual budget allocation.

Article 12: Financial and Incentive Measures

Article Content: Article 12 requires the government to actively promote AI R&D, applications, and infrastructure, properly plan resource allocation, and may adopt subsidies, commissioning, investment, rewards, guidance and other measures for AI-related industries, or provide tax and financial incentives.

Conceptual Analysis: This article authorizes the government to use various policy tools to support AI industry and technology development. Unlike Article 11’s focus on “having money to use,” Article 12 focuses on “how to use the money.”

Article 13: Regulatory Adaptation and Emerging Technology Priority

Article Content: Article 13 stipulates that when the government develops, trains, tests, and verifies emerging technology operational impacts, it should provide reasonable use, support, and subsidy measures, and improve AI R&D application regulations. When relevant regulations’ interpretation and application conflict with other laws, while conforming to Article 4’s basic principles, priority should be given to protecting emerging technology and service provision.

Conceptual Analysis: This article can be called the most innovative and breakthrough-spirited article in Taiwan’s AI Basic Act, demonstrating a “sandbox spirit” different from traditional regulatory thinking. Its core philosophy is that laws should adapt to technological evolution rather than letting innovation be constrained by old regulations.

Article 14: Public-Private Partnership for Innovation

Article Content: Article 14 requires the government to actively jointly promote AI innovation applications with the private sector based on public-private partnership principles.

Conceptual Analysis: Though only one sentence, this article highlights the key strategy for AI development: joining forces with industry and private sector. Public and private sectors each have advantages—government has resources, policies, and public data; enterprises have technology, creativity, and market sensitivity.

Article 15: Data Value-based Opening and Sharing Mechanisms

Article Content: Article 15 stipulates that based on data characteristics of permanence, encrypted reusability, and value traceability, the government should establish data opening, sharing, and reuse mechanisms, enhance AI data usability, and regularly review and adjust relevant legal norms. The government should strive to improve the quality and quantity of AI data used in Taiwan, ensuring training results adequately demonstrate national multicultural values and protect intellectual property rights.

Conceptual Analysis: This article is an extremely bright spot in the draft, called the “data value-based opening” clause, directly targeting AI development’s lifeline: data. AI training requires massive and diverse datasets; the more abundant and higher quality the data, the more accurate and useful AI models become.

Article 16: Personal Data Protection and Privacy by Design

Article Content: Article 16 requires the personal data protection authority (future Personal Data Protection Committee) to assist various competent authorities in avoiding unnecessary personal data collection, processing, or use during AI R&D and applications, and promote incorporating personal data protection into default design measures or mechanisms.

Conceptual Analysis: This article integrates privacy protection into the entire AI development process, implementing Article 4’s “privacy and data governance” principles. It emphasizes data minimization principles and Privacy by Design requirements.

Article 17: Labor Rights and Transition Measures

Article Content: Article 17 has three clauses: first, the government should actively use AI to ensure workers’ labor rights; second, the government should proactively bridge skill gaps caused by AI development; third, the government should provide employment guidance for those unemployed due to AI utilization.

Conceptual Analysis: This article proposes comprehensive countermeasures for AI’s impact on the job market, embodying the value that “technological progress should not come at workers’ expense.” While AI can improve productivity, it also raises concerns about “machines replacing human labor.”

Article 18: Risk Classification Framework and International Integration

Article Content: Article 18 requires the Ministry of Digital Affairs to coordinate relevant agencies, reference international standards or norms, promote internationally compatible AI risk classification frameworks, and conduct regular reviews. Various competent authorities may formulate risk classification norms for their supervised businesses based on this framework.

Conceptual Analysis: This article introduces risk-based management concepts, marking Taiwan’s AI governance move toward scientific and refined approaches. Risk classification ideas originate from the EU AI Act, which categorizes risks as “unacceptable, high-risk, limited risk, minimal risk” based on AI applications.

Article 19: Trustworthy AI Responsibility and Exemptions

Article Content: Article 19 first requires the government to enhance AI application trustworthiness through standards, verification, testing, marking, disclosure, traceability, and accountability mechanisms based on AI risk classification. However, AI technology development and research activities before application, except for following Article 4’s basic principles, are not subject to the aforementioned application responsibility regulations to facilitate technological innovation.

Conceptual Analysis: This article can be understood in two parts: during/post-application responsibility regulations and R&D stage responsibility exemptions. The exemption provides a safe sandbox for researchers and developers, allowing them not to worry about innovations being halted for failing to meet cumbersome standards before market release.

Article 20: Regulatory and Agency Function Adjustment Implementation

Article Content: Article 20 requires that after this law’s implementation, the government should review and adjust supervised functions, operations, and regulations according to this law to implement its purposes. Before relevant regulations are formulated or amended, if existing regulations have no provisions, central competent authorities shall jointly interpret and apply this law with the central authority (NSTC).

Conceptual Analysis: This article ensures the AI Basic Act doesn’t become a paper declaration but truly changes government operations through follow-up supporting measures. It provides transitional arrangements while new laws and systems are incomplete.

Article 21: Effective Date

Article Content: Article 21 stipulates this law takes effect from its promulgation date.

Conceptual Analysis: Though this article has no special technical content, “effective from promulgation” means the law can immediately take effect after Legislative Yuan passage and presidential announcement, showing legislators’ urgency and determination.

Conclusion

Examining each article of the Artificial Intelligence Basic Act draft reveals careful legislative design balancing “promoting development” and “addressing risks”: the first half emphasizes development orientation and ethical principles, while the second half constructs supporting measures and governance frameworks, complementing each other.

In international comparison, Taiwan’s basic law version aligns ideologically with innovation-friendly models like the US and Japan (encouraging experimentation, self-regulation guidance) while referencing EU risk regulation elements (classification, marking, accountability), integrating strengths from various approaches to create a balanced path suited to Taiwan’s national conditions.

This AI charter written for the next 10 years will lead Taiwan to catch up in international AI competition, injecting new momentum into the economy and bringing new welfare to society. Only through cooperative implementation by the Legislative Yuan and various government departments can Taiwan demonstrate its ability to give the world “Taiwan Speed” through law, standing out in the AI era!

📥 Related Resource Downloads

📄 AI Basic Act Original Documents

📖 Detailed Analysis Documents