Draft of AI Product and System Evaluation Guidelines Released by the Administration for Digital Industries to Enhance AI Governance
Draft of AI Product and System Evaluation Guidelines Released by the Administration for Digital Industries to Enhance AI Governance
2024/08/15
I. AI Taiwan Action Plan 2.0
In 2018, the Executive Yuan launched the “AI Taiwan Action Plan” to ensure that the country keeps pace with AI developments. This strategic initiative focuses on attracting top talent, advancing research and development, and integrating AI into critical sectors such as smart manufacturing and healthcare. The action plan has sparked growing discussion on AI regulation. Through these efforts, Taiwan aims to position itself as a frontrunner in the global smart technology landscape. Later in 2023, the Executive Yuan updated the action plan, releasing “AI Taiwan Action Plan 2.0” to further strengthen AI development.
“AI Taiwan Action Plan 2.0” outlines five main pillars:
1. Talent Development: Enhancing the quality and quantity of AI expertise, while improving public AI literacy through targeted education and training initiatives.
2. Technological and Industrial Advancement: Focusing on critical AI technologies and applications to foster industrial growth; and creating the Trustworthy AI Dialogue Engine (TAIDE) that communicates in Traditional Chinese.
3. Enhancing work environments: Establishing robust AI governance infrastructure to facilitate industry and governmental regulation, and to foster compliance with international standards.
4. International Collaboration: Expanding Taiwan's role in international AI forums, such as the Global Partnership on AI, to collaborate on developing trustworthy AI practices.
5. Societal and Humanitarian Engagement: Utilizing AI to tackle pressing societal challenges such as labor shortages, an aging population, and environmental sustainability.
II. AI Product and System Evaluation Guidelines: A Risk-based Approach to AI Governance
To support infrastructure, in March 2024, the Administration for Digital Industries issued the draft AI Product and System Evaluation Guidelines. The Guidelines are intended to serve as a reference for industry when developing and using AI products and systems, thus laying a crucial foundation for advancing AI-related policies in Taiwan. The Guidelines outline several potential risks associated with AI:
1. Third-Party Software and Hardware: While third-party software, hardware, and datasets can accelerate development, they may also introduce risks into AI products and systems. Therefore, effective risk management policies are crucial.
2. System Transparency: The lack of transparency in AI products and systems makes risk assessment relatively challenging. Inadequate transparency in AI models and datasets also pose risks for development and deployment.
3. Differences in Risk Perception: Developers of AI products and systems may overlook risks specific to different application scenarios. Moreover, risks may gradually emerge as the product or system is used and trained over time.
4. Application Domain Risks: Variations between testing results and actual operational performance can lead to differing risk assessments for evaluated products and systems.
5. Deviation from Human Behavioral Norms: If AI products and systems behave unexpectedly compared to human operations, this can indicate a drift in the product, system, or model, thereby introducing risks.
The Guidelines also specify that businesses have to categorize risks when developing or using AI products and systems, and manage them in accordance with these classifications. In alignment with the EU AI Act, risks are classified into four levels: unacceptable, high, limited, and minimal.
1. Unacceptable Risk: If AI systems used by public or private entities provide social scoring of individuals, this could lead to discriminatory outcomes and the exclusion of certain groups. Furthermore, if AI systems are employed to manipulate the cognitive behavior of individuals or vulnerable populations, causing physical or psychological harm, such systems are deemed unacceptable and prohibited.
2. High risk: AI systems are classified as high-risk in several situations. These include applications used in critical infrastructure, such as transportation, where there is potential risk to citizens' safety and health. These situations also encompass AI systems involved in educational or vocational training (such as exam scoring), which can determine access to education or professional paths. AI used as safety-critical product components, such as robot-assisted surgery, also falls into this category. In the employment sector, AI systems used for managing recruitment processes, including CV-sorting software, are considered high-risk. Essential private and public services, such as credit scoring systems that impact loan eligibility, also fall under high-risk. AI used in law enforcement in ways that it may affect fundamental rights, such as evaluating the reliability of evidence, is also included. AI systems involved in migration, asylum, and border control, such as automated visa application examinations, are categorized as high-risk. Finally, AI solutions used in the administration of justice and democratic processes, such as court ruling searches, are also classified as high-risk.
If an AI system is classified as high risk, it must be evaluated across ten criteria—Safety, Explainability, Resilience, Fairness, Accuracy, Transparency, Accountability, Reliability, Privacy, and Security—to ensure the AI system’s quality.
3. Limited risk: When an AI product or system is classified as having limited risk, it is up to the enterprise to determine whether an evaluation is required. The Guidelines also introduce specific transparency obligations to ensure that humans are informed when necessary, thus fostering trust. For instance, when using AI systems such as chatbots or systems for generating deepfake content, humans must be made aware that they are interacting with a machine so they can take an informed decision to continue or step back.
4. Minimal or no risk: The Guidelines allow the free use of minimal-risk AI. This includes applications such as AI-enabled video games and spam filters.
Ⅲ. Conclusion
The AI Product and System Evaluation Guidelines represent a significant step forward in establishing a robust, risk-based framework for AI governance in Taiwan. By aligning with international standards like the EU AI Act, these Guidelines ensure that AI products and systems are rigorously assessed and categorized into four distinct risk levels: unacceptable, high, limited, and minimal. This structured approach allows businesses to manage AI-related risks more effectively, ensuring that systems are safe, transparent, and accountable.
The emphasis on evaluating AI systems across ten critical criteria—including safety, explainability, and fairness—reflects a comprehensive strategy to mitigate potential risks. This proactive approach not only safeguards the public but also fosters trust in AI technologies. By setting clear expectations and responsibilities for businesses, the Guidelines promote responsible development and deployment of AI, ultimately contributing to Taiwan's goal of becoming a leader in the global AI landscape.
Brief Introduction to Taiwan Social Innovation Policies 2021/09/13 1. Introduction The Millennium Development Goals (MDGs)[1] set forth by the United Nations in 2000 are carried out primarily by nations and international organizations. Subsequently, the Sustainable Development Goals (SDGs) set forth by the United Nations in 2015 started to delegate the functions to organizations of all levels. Presently, there is a global awareness of the importance of balancing “economic growth”, “social progress”, and “environmental protection” simultaneously during development. In the above context, many similar concepts have arisen worldwide, including social/solidarity economy, social entrepreneurship and social enterprise, and social innovation. Generally, social innovation aims to alter the interactions between various groups in society through innovative applications of technology or business models, and to find new ways to solve social problems through such alterations. In other words, the goal is to use innovative methods to solve social problems.The difference between social innovation and social enterprise is that social enterprise combines commercial power to achieve its social mission under a specific perspective, while social innovation creates social value through cooperation with and coordination among technology, resources, and communities under a diversified nature. 2. Overview of Taiwan Social Enterprise Policy To integrate into the global community and assist in the development of domestic social innovation, Taiwan’s Executive Yuan launched the “Social Enterprise Action Plan” in 2014, which is the first policy initiative to support social enterprises (from 2014 to 2016).Under this policy initiative, through consulting with various ministries and applying methods such as “amending regulations”, “building platforms”, and “raising funds”, the initiative set to create an environment with favorable conditions for social innovation and start-ups. At this stage, the initiative was adopted under the principle of “administrative guidance before legislation” in order to encourage private enterprise development without excessive burden, and avoid regulations restricting the development of social enterprises, such as excessive definition of social enterprises. Moreover, for preserving the original types of these enterprises, this Action Plan did not limit the types of social enterprises to companies, non-profit organizations, or other specific types of organizations. To sustain the purpose of the Social Enterprise Action Plan and to echo and reflect the 17 sustainable development goals proposed in SDGs by the United Nations, the Executive Yuan launched the “Social Innovation Action Plan” (effective from 2018 to 2022) in 2018 to establish a friendly development environment for social innovation and to develop diversified social innovation models through the concept of “openness, gathering, practicality, and sustainability”.In this Action Plan, “social innovation” referred to “social innovation organizations” that solve social problems through technology or innovative business models. The balancing of the three managerial goals of society, environment value, and profitability is the best demonstration of the concept of social innovation. 3. Government’s Relevant Social Enterprise Policy and Resources The ministries of the Taiwan Government have been promoting relevant policies in accordance with the Social Innovation Action Plan issued by the Executive Yuan in 2018, such as the “Registration System for Social Innovation Enterprises” (counseling of social enterprises), the “Buying Power - Social Innovation Products and Services Procurement”, the “Social Innovation Platform” established by the Ministry of Economic Affairs, the “Social Innovation Manager Training Courses”, the “Promoting Social Innovation and Employment Opportunities” administered by the Ministry of Labor, and the “University Social Responsibility Program” published by the Ministry of Education. Among the above policies stands out the measures adopted by the Ministry of Economic Affairs, and a brief introduction of those policies are as follows: i. Social Innovation Platform To connect all resources involved in social issues to promote social innovation development in Taiwan, the Ministry of Economic Affairs established the “Social Innovation Platform”.[2] With visibility through the Social Innovation Platform, it has become more efficient to search for targets in a public and transparent way and to assist with the input of resources originally belonging to different fields in order to expand social influence. As a digital platform gathering “social innovation issues in Taiwan,” the Social Innovation Platform covers multiple and complete social innovation resources, which include the “SDGs Map” constructed on the Social Innovation Platform, by which we can better understand how county and city governments in Taiwan implement SDGs and Voluntary Local Review Reports, and which allow us to search the Social Innovation Database[3] and the registered organizations, by which citizens, enterprises, organizations, and even local governments concerned with local development can find their partners expediently as possible, establish service lines to proactively assist public or private entities with their needs/resources, and continue to enable the regional revitalization organizations, ministries, and enterprises to identify and put forward their needs for social innovation through the function of “Social Innovation Proposals”, which assist social innovation organizations with visibility while advancing cooperation and expanding social influence. In addition, the “Event Page” was established on the Social Innovation Platform and offers functions, such as the publishing, searching, and sorting of events in four major dimensions with respect to social innovation organization, governments, enterprises, and citizens; and encourages citizens, social innovation organizations, enterprises, and governments to devote themselves via open participation to continuously expande the influence of the (Civic Technology) Social Innovation Platform. The “Corporate Social Responsibility Report” collects the corporate social responsibility reports, observes the distribution of resources for sustainable development by corporations in Taiwan, offers filtering functions by regions, keyword, popular rankings, and or SDGs types, and provides contact information and a download function for previous years’ reports, in order to effectively assist social innovation organizations to obtain a more precise understanding of the status quo, needs, and trends with respect to their development of respective products and services. Figure 1: SDGs Map Reference: Social Innovation Platform (https://si.taiwan.gov.tw/) Figure 2: Social Innovation Database Reference: Social Innovation Platform (https://si.taiwan.gov.tw/) Figure 3: Social Innovation Proposals Reference: Social Innovation Platform (https://si.taiwan.gov.tw/) Figure 4: Event Page Reference: Social Innovation Platform (https://si.taiwan.gov.tw/) Figure 5: Corporate Social Responsibility Report Reference: Social Innovation Platform (https://si.taiwan.gov.tw/) ii. Social Innovation Database To encourage social innovation organizations to disclose their social missions, products and services, and to guide society to understand the content of social innovation, and to assist the administrative ministries to be able to utilize such information, the Ministry of Economic Affairs issued the “Principles of Registration of Social Innovation Organizations” to establish the “Social Innovation Database”. Once a social innovation organization discloses the items, such as its social missions, business model, or social influence, it may obtain the relevant promotional assistance resources, including becoming a trade partner with Buying Power (Social Innovation Products and Services Procurement), receiving exclusive consultation and assistance from professionals for social innovation organizations, and becoming qualified to apply to entering into the Social Innovation Lab.Moreover, the Ministry of Economic Affairs is simultaneously consolidating, identifying, and designating the awards and grants offered by the various ministries, policies and measures in respect of investment, and financing and assistance, as resources made available to registered entities. As of 25 May 2021, there were 658 registered social innovation organizations and 96 Social Innovation Partners (enterprises with CSR or ESG resources that recognize the cooperation with social innovation under the social innovation thinking model may be registered as a “Social Innovation Partner”).The public and enterprises can search for organizations registered in the Social Innovation Database through the above-said Social Innovation Platform, the search ability of which advances the exposure of and the opportunities for cooperation with social innovation organizations. Figure 6: Numbers of registered social innovation organizations and accumulated value of purchases under Buying Power Reference: Social Innovation Platform(https://si.taiwan.gov.tw/) iii. Buying Power - Social Innovation Products and Services Procurement In order to continue increasing the awareness on social innovation organizations and related issues and promote responsible consumption and production in Taiwan, as well as to raise the attention of the commercial sector to the sustainability-driven procurement models, the Ministry of Economic Affairs held the first “Buying Power - Social Innovation Products and Services Procurement” event in 2017. Through the award system under the Buying Power, it continues to encourage the governments, state-owned enterprises, private enterprises, and organizations to take the lead in purchasing products or services from social innovation organizations, to provide the relevant resources so as to assist social innovation organizations to obtain resources and to explore business opportunities in the markets, to practice responsible consumption and production, and to promote innovative cooperation between all industries and commerce and social innovation organizations. The aim of the implementation of the Buying Power is to encourage the central and local governments, state-owned enterprises, private enterprises, and non-governmental organizations to purchase products or services from organizations registered in the Social Innovation Database, while prizes will be awarded based on the purchase amounts accumulated during the calculation period. The winners can obtain priority in applying for membership in the Social Innovation Partner Group, with corresponding member services, in the future. Under the Social Innovation Platform, both the amount of purchase awards and the number of applicants for special awards continue to increase.So far, purchases have accumulated to a value of more than NT$1.1 billion (see Figure 6), and more than 300 organizations have proactively participated. iv. Social Innovation Mark In order to promote public awareness of social innovation, the Ministry of Economic Affairs has been charged with the commissioned task of promoting the Social Innovation Mark, and issued “ The Small and Medium Enterprise Administration of the Ministry of Economic Affairs Directions for Authorization of the Social Innovation Mark” as the standard for the authorization of the Social Innovation Mark. Social innovation organizations can use the Mark, through obtaining authorization, to hold Social Innovation Summits or other social innovation activities for promoting social innovation concepts. In order to build the Mark as a conceptual symbol of social innovation, the Ministry of Economic Affairs has been using the Social Innovation Mark in connection with various social innovation activities, such as the Social Innovation Platform, the Buying Power, and the annual Social Innovation Summit. Taking the selection of sponsors of the Social Innovation Summit in 2022 as an example[4], only organizations that have obtained authorization of the Social Innovation Mark can use the Mark to hold the Social Innovation Summit. Figure 7: The Social Innovation Mark of the Small and Medium Enterprise Administration, Ministry of Economic Affairs IV. Conclusion The “Organization for Economic Cooperation and Development” (OECD) regards social innovation as a new strategy for solving future social problems and as an important method for youth entrepreneurship and social enterprise development.Taiwan’s social innovation energy has entered a stage of expansion and development. Through the promotion of the “Social Innovation Action Plan,” the resources from the central and local governments are integrated to establish the Social Innovation Platform, the Social Innovation Database, the Social Innovation Lab, and the Social Innovation Mark. In addition, incentives such as the Buying Power have been created, manifesting the positive influence of Taiwan’s social innovation. [1] MDGs are put forward by the United Nations in 2000, and are also the goals requiring all the 191 member states and at least 22 international organizations of the United Nations to be committed to on their best endeavors, including: 1. eradicating extreme poverty and hunger, 2. applying universal primary education, 3. promoting gender equality and empowering women, 4. reducing child mortality rates, 5. improving maternal health, 6. combatting HIV/AIDS, malaria, and other diseases, 7. ensuring environmental sustainability, and 8. establishing a global partnership for development. [2] Please refer to the Social Innovation Platform: https://si.taiwan.gov.tw/. [3] Please refer to the Social Innovation Database: https://si.taiwan.gov.tw/Home/Org_list. [4] Please refer to the guidelines for the selection of sponsors of the 2022 Social Innovation Summit: https://www.moeasmea.gov.tw/files/6221/4753E497-B422-4303-A8D4-35AE0B4043A9
A Before and After Impact Comparison of Applying Statute for Industrial Innovation Article 23-1 Draft on Venture Capital Limited PartnershipsA Before and After Impact Comparison of Applying Statute for Industrial Innovation Article 23-1 Draft on Venture Capital Limited Partnerships I. Background Because the business models adopted by Industries, such as venture capital, film, stage performance and others, are intended to be temporary entities, and the existing business laws are not applicable for such industries,[1] the Legislature Yuan passed the “Limited Partnership Act” in June 2015,[2] for the purpose of encouraging capital injection into these industries. However, since the Act was passed, there are currently only nine limited partnerships listed on the Ministry of Economic Affairs' limited partnership information website. Among them, “Da-Zuo Limited Partnership (Germany) Taiwan Branch” and “Stober Antriebstechnik Limited Partnership (Germany) Taiwan Branch”, are branch companies established by foreign businesses, the remaining seven companies are audio video production and information service businesses. It is a pity that no venture capital company is adopting this format.[3] In fact, several foreign countries have set up supporting measures for their taxation systems targeting those business structures, such as limited partnerships. For example, the pass-through taxation method (or referred to as single entity taxation) is adopted by the United States, while Transparenzprinzip is used by Germany. These two taxation methods may have different names, but their core ideas are to pass the profits of a limited partnership to the earnings of partners.[4] However, following the adoption of the Limited Partnership Act in Taiwan, the Ministry of Finance issued an interpretation letter stating that because the current legal system confers an independent legal entity status to the business structure of a limited partnership, it should be treated as a profit-seeking business and taxed with Profit-Seeking Enterprise Income Tax.[5] Therefore, to actualize the legislative objective of encouraging innovative businesses organized under tenets of the Limited Partnership Act, the Executive Yuan presented a draft amendment for Article 23-1 of the Statute for Industrial Innovation (hereinafter referred to as the Draft), introducing the "Pass Through Taxation Principle" as adopted by several foreign countries. That is, a Limited Partnership will not be levied with the Profit-Seeking Enterprise Income Tax, but each partner will file income tax reports based on after-profit-gains from the partnership that are passed through to each partner. It is expected that the venture capital industry will now be encouraged to adopt the limited partnership structure, and thus increase investment capital in new ventures. II. The Pass Through Taxation Method is Applicable to Newly Established Venture Capital Limited Partnerships 1. The Requirements and Effects (1) The Requirements According to the provisions of Article 23-1 Paragraph 3 of the Draft, to be eligible for Pass Through Taxation, newly established venture capital limited partnerships must meet the following requirements: 1. The venture capital limited partnerships are established between January 1, 2017 and December 31, 2019. 2. Investment threshold of the total agreed capital contribution, total received capital contribution, and accumulated total capital contribution, within five years of the establishment of venture capital limited partnerships: Total Agreed Capital Contribution in the Limited Partnership Agreement Total Received Capital Contribution Accumulated Investment Amount for Start-up Companies The Year of Establishment 3 hundred million ✕ ✕ The Second Year ✕ ✕ The Third Year 1 hundred million ✕ The Fourth Year 2 hundred million Reaching 30 percent of the total received capital contribution of the year or 3 hundred million NT dollars. The Fifth Year 3 hundred million 3. The total amount, that an overseas company applies in capital and investments in actual business operations in Taiwan, reaches 50% of its total received capital contribution of that year. 4. In compliance with government policies. 5. Reviewed and approved by the central competent authority each year. (2) The Effects The effects of applying the provisions of Article 23-1 Paragraph 3 of the Draft are as follows: 1. Venture capital limited partnerships are exempt from the Profit-Seeking Enterprise Income Tax. 2. Taxation method for partners in a limited partnership after obtaining profit gains: (1) Pursuant to the Income Tax Act, Individual partners and for-profit business partners are taxed on their proportionally-calculated, distributed earnings. (2) Individual partners and foreign for-profit business partners are exempt from income tax on the stock earnings distributed by a limited partnership. 2. Benefit Analysis Before and After Applying Pass Through Taxation Method A domestic individual A, a domestic profit-making business B, and a foreign profit-making business C jointly form a venture capital limited partnership, One. The earnings distribution of the company One is 10%, 80% and 10% for A, B, and C partners, respectively. The calculated earnings of company One are one million (where eight hundred thousand are stock earnings, and two hundred thousand are non-stock earnings). How much income tax should be paid by the company One, and partners A, B, and C? (1) Pursuant to the Income Tax Act, before the amended draft: 1. One Venture Capital Limited Partnership Should pay Profit-Seeking Enterprise Income Tax = (NT$1,000,000 (earning) - NT$500,000[6])x12% (tax rate[7])=NT$60,000 2. Domestic Individual A Should file a comprehensive income report with business profit income =(NT$1,000,000-NT$60,000) x 10% (company One draft a voucher for net amount for A) + NT$60,000÷2×10% (deductible tax rate)= NT$97,000 Tax payable on profit earnings＝NT$91,500×5%(tax rate)＝NT$4,850 Actual income tax paid＝NT$4,850 - NT$60,000÷2×10% (deductible tax rate) ＝NT$1,485 3. Domestic For-Profit Business B Pursuant to the provisions of Article 42 of the Income Tax Act, the net dividend or net income received by a profit-seeking company is not included in the income tax calculation. 4. Foreign For-Profit Business C Tax paid at its earning source＝(NT$1,000,000 - NT$60,000) ×10% (earning distribution rate) ×20% (tax rate at earning source)＝NT$18,800 (2) Applying Pass Through Taxation Method After Enacting the Amendment 1. One Venture Capital Limited Partnership No income tax. 2. Domestic Individual A Should pay tax＝NT$800,000 (non-stock distributed earnings)×10% (earning distribution rate)×5% (comprehensive income tax rate)＝NT$1,000 3. Domestic For-Profit Business B Pursuant to the provisions of Article 42 of the Income Tax Act, the net dividend or net income received by a profit-seeking company is not included in the income tax calculation. 4. Foreign For-Profit Business C Tax paid at its earning source＝NT$800,000 (non-stock distributed earnings)×10%(earning distribution rate)×20% (tax rate at earning source)＝NT$4,000 The aforementioned example shows that under the situation, where the earning distribution is the same and tax rate for the same taxation subject is the same, the newly-established venture capital limited partnerships and their shareholders enjoy a more favorable tax benefit with the adoption of pass through taxation method: Before the Amendment After the Amendment Venture Capital Limited Partnership NT$60,000 Excluded in calculation Shareholders Domestic Individual NT$1,850 NT$1,000 Domestic For-Profit Business Excluded in calculation Excluded in calculation Foreign For-Profit Business NT$18,800 NT$4,000 Sub-total NT$80,650 NT$5,000 III. Conclusion Compared to the corporate taxation, the application of the pass through taxation method allows for a significant reduction in tax burden. While developing Taiwan’s pass through tax scheme, the government referenced corporate taxation under the U.S. Internal Revenue Code (IRC), where companies that meet the conditions of Chapter S can adopt the “pass through” method, that is, pass the earnings to the owner, with the income of shareholders being the objects of taxation;[8] and studied the "Transparenzprinzip" adopted by the German taxation board for partnership style for-profit businesses. Following these legislative examples, where profits are identified as belonging to organization members,[9] the government legislation includes the adoption of the pass through taxation scheme for venture capital limited partnerships in the amended draft of Article 23-1 of the Statute for Industrial Innovation, so that the legislation is up to international standards and norms, while making an important breakthrough in the current income tax system. This is truly worthy of praise. [1] The Legislative Yuan Gazette, Vol. 104, No. 51, page 325. URL:http://misq.ly.gov.tw/MISQ//IQuery/misq5000Action.action [2] A View on the Limited Partnership in Taiwan, Cross-Strait Law Review, No. 54, Liao, Da-Ying, Page 42. [3] Ministry of Economic Affairs - Limited Partnership Registration Information URL: http://gcis.nat.gov.tw/lmpub/lms/dir.jsp?showgcislocation=true&agencycode=allbf [4] Same as annotate 2, pages 51-52. [5] Reference Letter of Interpretation dated December 18, 2015, Tai-Cai-Shui Zi No. 10400636640, the Ministry of Finance [6] First half of Paragraph 1 of Article 8 of the Income Basic Tax Act [7] Second half of Paragraph 1 of Article 8 of the Income Basic Tax Act [8] A Study on the Limited Partnership Act, Master’s degree thesis, College of Law, Soochow University, Wu, Tsung-Yeh, pages 95-96. [9] Reference annotate 2, pages 52.
Taiwan’s AI Risk Taxonomy and Assessment Framework: A Risk Governance Approach under the Artificial Intelligence Basic ActTaiwan’s AI Risk Taxonomy and Assessment Framework: A Risk Governance Approach under the Artificial Intelligence Basic Act 2026/09/30 I. Introduction As artificial intelligence technologies advance rapidly and spread across industries, the focus of AI governance worldwide has shifted from declaring ethical principles to building concrete rules. On January 14, 2026, Taiwan promulgated the Artificial Intelligence Basic Act (the “AI Basic Act”), which took effect on the same day and serves as the legal foundation for AI governance in Taiwan. Pursuant to Article 16 of the Act, the Ministry of Digital Affairs (MODA) developed the AI Risk Taxonomy and Assessment Framework (the “Framework”). After approval by the Executive Yuan, MODA released Version 1.0 of the Framework on July 7, 2026. Notably, AI in Taiwan concerns not only economic and industrial development but also national security, and this consideration is reflected in how the Framework’s risk types are designed. This article focuses on the Framework. It explains the structure of the AI Basic Act, Taiwan’s AI risk governance system, and the Framework’s positioning and operational workflow, and it briefly compares Taiwan’s approach with those of other major jurisdictions. The aim is to help the international community understand the direction of AI risk governance in Taiwan. II. The Two Pillars of the AI Basic Act The AI Basic Act aims to promote innovation while safeguarding human rights and human autonomy. Broadly, the Act rests on two pillars. The first is promoting development, which covers industrial innovation, data governance, and talent cultivation. The second is risk management, which covers the Framework, warnings for high-risk applications, and the attribution of liability. In terms of its legal nature, the AI Basic Act can be understood as a national AI strategy established in the form of law. It applies to government agencies and does not directly regulate businesses. In other words, the specific obligations of businesses continue to come from regulations that the competent authorities for the relevant industries (the “competent authorities”) adopt within their respective areas of responsibility. Competent authorities are the central government agencies that, by law, oversee particular industries or affairs. For example, the Ministry of Health and Welfare oversees healthcare, and the Financial Supervisory Commission oversees financial services. On the risk management side, the relevant provisions are as follows. Article 5, Paragraph 1 requires the government to prevent AI applications from infringing on people’s life, body, freedom, or property, or from disrupting social order, national security, or the ecological environment. Under Article 16, MODA promotes an AI Risk Taxonomy and Assessment Framework, and the competent authorities establish management regulations in accordance with that framework. Article 17 requires the government to clarify the attribution of liability and the conditions for such liability for high-risk AI applications, and to establish relief, compensation, or insurance mechanisms. Article 18 requires government agencies to complete the review and adaptation of laws and regulations by January 13, 2028. III. A Three-Tier Risk Governance Structure Taiwan’s AI risk governance system can be understood in three tiers. The first tier consists of development principles. Article 4 of the AI Basic Act sets out seven principles: sustainable development and well-being; human autonomy; privacy protection and data governance; cybersecurity and safety; transparency and explainability; fairness and non-discrimination; and accountability. These principles are similar to the AI Principles of the Organisation for Economic Co-operation and Development (OECD) and form the foundation of the overall governance system. The second tier is the Framework. Adopted under Article 16, Paragraph 1, the Framework groups AI risks into three types: technical design flaws in AI systems; issues in post-deployment operation and human-AI interaction; and societal-structural and environmental impacts. The third tier consists of the governance work of individual ministries and agencies. Each of the three risk types is matched with a different governance tool. Technical design flaws are addressed through assessment and verification tools. Issues in post-deployment operation and human-AI interaction are addressed through regulatory measures. Societal-structural and environmental impacts are addressed through government policies. In short, technology is checked through assessment and verification, use is governed by regulation, and broad social impacts are guided by policy. This design matches governance tools to the nature of each risk and avoids handling every risk through regulation. IV. Positioning and Design of the Framework First, the Framework serves the competent authorities. It helps them establish risk-based regulations for AI applications in the industries under their purview. “Risk-based” means that the higher the risk, the stricter the rules, while lower-risk applications retain room for innovation. Second, the Framework is positioned as a methodological tool. It guides the competent authorities and does not directly regulate businesses. In this respect, it is similar to the AI Risk Management Framework (AI RMF) of the U.S. National Institute of Standards and Technology (NIST). Both are process-oriented, and neither method is in itself a form of regulation. Their legal status, however, differs. The NIST AI RMF is adopted by businesses on a voluntary basis, whereas the Framework has a statutory basis and binds the procedures that government agencies follow in developing regulations. Third, the Framework neither relies on a list of high-risk AI predefined in law nor sorts AI into fixed risk levels. Instead, it uses specific use cases as the unit of analysis, and the competent authorities determine whether a use case is high-risk and state their reasons. This approach reflects three considerations: the same technology can pose very different risks in different contexts; technology changes rapidly; and each authority knows the industries under its purview best. To prevent inconsistent standards across agencies, the Framework also provides three common tools: a unified Table of AI Risk Types, a unified set of protected legal interests for determining serious harm, and a Table of Illustrative Examples of Serious Harm. In addition, the Framework is centered on promoting innovation. When planning measures, agencies should follow the principle of proportionality and assess the opportunity costs of not adopting AI, so that an overly cautious approach does not leave policy lagging behind social needs. The Framework applies to the civilian domain and excludes military uses. V. The Four-Step Operational Workflow The Framework’s operational workflow consists of four steps, described below in order. Step 1: AI Use Case Inventory Each agency takes inventory of existing and planned AI use cases in the industries under its purview and records their purposes, technologies, and stakeholders. Step 2: Risk Identification Each agency checks each use case against the Table of AI Risk Types, item by item, to identify the risks involved. The table sets out 20 subtypes under three categories: Risk Type Subtypes A: Technical Design Flaws in AI Systems A1 AI System Security Vulnerabilities and Attacks A2 Lack of Transparency or Explainability A3 AI Behavior Misaligned with Human Intent and Social Values A4 Dangerous Capabilities of AI Systems A5 Privacy Infringement and Non-Compliance with Personal Data Protection Laws A6 Risk of Intellectual Property Infringement A7 Unfair Discrimination or Bias A8 False or Misleading Information B: Issues in Post-Deployment Operation and Human-AI Interaction B1 Overreliance and Unsafe Use B2 Loss of Human Autonomy B3 Generation of Unlawful Content B4 Fraud and Misuse of Deepfake Technology B5 Facilitation of Cyberattacks B6 Unauthorized Actions by Autonomous AI Agents C: Societal-Structural and Environmental Impacts C1 Competitive Imbalance Among Companies and Countries C2 Power Centralization and Unfair Distribution of Benefits C3 Increased Inequality and Decline in Employment Quality C4 Economic and Cultural Devaluation of Human Creativity C5 Environmental Harm C6 Cognitive Warfare and Information Sovereignty Among these, C6 (Cognitive Warfare and Information Sovereignty) addresses the fact that AI greatly lowers the cost of large-scale information manipulation. Malicious actors can use generative AI, automated accounts, and deepfake content to systematically interfere with public discourse, electoral processes, and policymaking. Few jurisdictions include this item in an official risk taxonomy, and its inclusion reflects Taiwan’s geopolitical environment. Step 3: Risk Assessment The assessment is based on the objective characteristics of the use case itself and does not take into account the mitigating effects of existing measures. A use case that may cause serious harm to any of six protected legal interests, namely national security, the fundamental rights of the people, life safety, property protection, social order, or the ecological environment, is a high-risk AI application. Step 4: Risk Response Each agency reviews how well the laws and regulations under its purview cover the identified risks and confirms whether they are sufficient to address them. If they are not, the agency plans measures in support of development or management measures in accordance with the principle of proportionality. For high-risk applications, the AI Basic Act requires that warnings be labeled and that liability be clearly attributed. Where a risk cannot be effectively controlled, the competent authority should restrict or prohibit the application in accordance with the law. Consider a hypothetical example from the healthcare sector. After the Ministry of Health and Welfare identifies AI-assisted diagnosis as a use case in its inventory, it identifies risks such as erroneous outputs and overreliance. Because a misdiagnosis could endanger lives, the use case is assessed as a high-risk AI application. The Ministry then reviews its healthcare regulations and requires that warnings be labeled and that liability be clearly attributed. If the existing regulations fall short, it adopts new rules. VI. Implementation Status To support implementation, MODA has built an online system for internal government use. Agencies use the system to report their inventory and assessment results, which allows MODA to track progress and provide timely assistance. MODA is also giving priority to working with ministries whose responsibilities directly affect the significant rights and interests of the public. It provides explanations and advice on applying the Framework, on risk concepts, and on international cases. Risk determinations and rulemaking, however, remain the responsibility of each agency within its own authority. The results of this work will serve as the basis for completing regulatory adaptation by January 13, 2028, and the Framework will be revised on a rolling basis in light of practical experience. VII. Comparative Observations Internationally, the European Union’s Artificial Intelligence Act (AI Act) defines high-risk AI systems in advance through Annexes I and III and imposes obligations on providers and deployers. However, the European Commission’s Digital Omnibus on AI, which entered into force in July 2026, postponed the application of high-risk obligations to December 2027 and August 2028. South Korea’s Framework Act on the Development of Artificial Intelligence and the Establishment of a Foundation for Trustworthiness took effect on January 22, 2026. It uses the concept of “high-impact AI” and lists the regulated areas in the statute itself. Japan enacted the Act on the Promotion of Research, Development and Utilization of Artificial Intelligence-Related Technologies in 2025. That Act focuses on promoting AI research, development, and utilization and contains no penalties. Compared with these systems, Taiwan’s approach has three distinctive features. First, high risk is determined at the level of specific use cases, and the competent authorities make case-by-case determinations and state their reasons. Second, risk governance follows the existing legal framework and is carried out through each agency’s review of its laws and regulations. Third, the Framework’s risk taxonomy draws on mainstream international norms, so multinational companies can continue to use their existing global compliance frameworks. VIII. Conclusion Overall, Taiwan is re-examining its laws and regulations across sectors on a risk basis while building on its existing regulatory structures, so that its legal system can respond flexibly to rapidly changing AI technologies. As agencies gain practical experience and international developments evolve, the Framework will continue to be reviewed and revised, with the aim of striking a balance between promoting innovation and protecting people’s rights and interests. References [1]Artificial Intelligence Basic Act (promulgated Jan. 14, 2026), Laws & Regulations Database of the Republic of China (Taiwan), https://law.moj.gov.tw/ENG/LawClass/LawAll.aspx?pcode=H0160093 . [2]Ministry of Digital Affairs, AI Risk Taxonomy and Assessment Framework, Version 1.0 (July 7, 2026), https://moda.gov.tw/en/major-policies/ai/governance/19244 .
Suggestions for MOEA Trial Program of Voluntary Base Green Electricity FrameworkOn March 6, 2014, The Energy Bureau of Ministry of Economic Affairs has published a pre-announcement on a Trial Program of Voluntary Base Green Electricity Framework （hereafter the Trial Program） and consulted on public opinion. In light of the content of the Trial Program, STLI provide the following suggestions for future planning of related policy structure. The institution of green electricity as established by the Trial Program is one of the policies for promoting renewable energy. Despite its nature of a trial, it is suggested that a policy design with a more options will be beneficial to the promotion of renewable energy, in light of various measures that have been undertaken by different countries. According to the Trial Program, the planned price rate of the green electricity is set on the basis of the total sum that the electricity subsidy to be paid by the Renewable Energy Development Fund divided by the total sum of electricity generated reported by Tai Power Company. The Ministry of Economic Affairs will adjust the price rate of the green electricity on the base of both how many users subscribe to the green electricity and the price rate of international green electricity market rate and, then announce the price rate in October of each year if not otherwise designated. In addition, according to the planned Trial Program, the unit for the subscription of green electricity is 100 kW·h. It is further reported that the current planned price rate for green electricity is 1.06 NTD/ kW·h. And it shall be 3.95 NTD/ kW·h if adding up with the original price rate, with an 37% increase in price per kW·h. In terms of the existing content of the Trial Program, only single price rate will be offered during the trial period. In this regard, we take the view that it would be beneficial to take into account similar approaches that have been taken by other countries. In Germany, for instance, the furtherance of renewable energy is achieved by the obligatory charge（EEG Umlage）together with the voluntary green electricity program provided by the private electricity retail sectors. According to German Ministry of Economics and Energy （BMWi）, the electricity price that the German public pays includes three parts: （1）the cost of the purchase and distribution of the electricity, including the margin of the electricity provider（2）regulated network fees, including those for the operation as well as for the measurement works of the meters（3）charges imposed by the government, including tax and the abovementioned obligatory charge for renewable energy（EEG Umlage）, as prescribed by the Act on Renewable Energy （Gesetz für den Vorrang Erneuerbarer Energien, also known as Erneuerbare-Energien-Gesetz - EEG）. In terms of how it is implemented on the ground, an example of the green electricity price menu program from the German electricity retail company, Vattenfall, is given in the following. In all price menu programs provided by Vattenfall in Berlin, for instance, 29.4% of the electricity comes from renewable energy as a result of the implementation of the Act on Renewable Energy. Asides from the abovementioned percentage as facilitated by the existing obligatory measures, the electricity retail companies in Germany further provide the price menus that are “greener”. For example, among the options provided by Vattenfall（Chart I）, in terms of the 12-month program, one can choose the menu which consist of 39.4% of renewable energy, with the price of 0.2642 Euro/ kW·h（about 10.96 NTD/ kW·h）. One can also opt for a menu of which the energy supply comes from 100% of renewable energy, with the price of 0.281 Euro/ kW·h（about 11.66 NTD/ kW·h） Chart I : Green Electricity Price Menus provided by Vattenfall in Berlin, Germany Percentage of Renewable Energy Supply Percentage of Renewable Energy Supply Electricity Price 12-month program 39.4% 0.2642 Euro/ kW·h（about 10.96 NTD/ kW·h） All renewable energy program 100% 0.281 Euro/ kW·h（about 11.66 NTD/ kW·h） Source：Vattenfall website, translated and reorganized by STLI, April 214. In addition, Australia also has similar programs on green electricity that is voluntary-base and with the goal of promoting renewable energy, reducing carbon emission, and transforming energy economy. Since 1997, the GreenPower in Australia is in charge of audition and certification of the retail companies and power plants on green electricity. The Australian model uses the certification mechanism conducted by independent third party, to ensure the green electricity purchased by end users in compliance with specific standards. As for the options for the price menu, take the programs of green electricity offered by the Australian retail company Origin Energy for example, user can choose 6 kinds of different programs, which are composed by renewable energy supply of respectively 10%, 20%, 25%, 50%, 75%, and 100%, at various price rates （shown in Chart II）. Chart II Australian Green Electricity Programs provided by Origin Energy Percentage of renewable Energy Electricity Price per kW·h 0 0.268 AUD（About 7.52 NTD） 10% 0.274868 AUD（About 7.69 NTD） 20% 0.28006 AUD（About 7.84 NTD） 25% 0.28292 AUD（About 7.92 NTD） 50% 0.2838 AUD（About 7.95 NTD） 100% 0.2992 AUD（About 8.37 NTD） Source：Origin Energy website, translated and reorganized by STLI, April 214. Given the information above, it can thus be inferred that the international mechanism for the promotion of green electricity often include a variety of price menus, providing the user more options. Such as two difference programs offered by Vattenfall in Germany and six various rates for green electricity offered by Origin Energy in Australia. It is the suggestion of present brief that the Trial Program can reference these international examples and try to offer the users a greater flexibility in choosing the most suitable programs for themselves.
- The amendment of the Taiwanese Personal Data Protection Act
- Impact of Government Organizational Reform to Scientific Research Legal System and Response Thereto (1) – For Example, The Finnish Innovation Fund (“SITRA”)
- Legal issues of Third-Party Payment in Taiwan
- The Demand of Intellectual Property Management for Taiwanese Enterprises
- Recent Federal Decisions and Emerging Trends in U.S. Defend Trade Secrets Act Litigation
- Blockchain in Intellectual Property Protection
- Impact of Government Organizational Reform to Research Legal System and Response Thereto (2) – Observation of the Swiss Research Innovation System
- Taiwan's Approach to AI Governance