Statement｜Moving Beyond Policy Declarations to Immediately Initiate Legislative Engineering for the Protection of the People: JRF Statement on the Legislative Yuan's Passage of the "Artificial Intelligence Basic Act"

[Corrected on March 17, 2026, as noted in the footnotes below]

Regarding the Artificial Intelligence Basic Act (hereinafter the "AI Basic Act") passed by the Legislative Yuan today, the Judicial Reform Foundation (JRF), as an organization that has long focused on the rule of law in the digital age, expresses grave concern that the Act fails to fully deliver on its commitment to a "human-centric" approach. Beyond vague development principles with little practical binding force, the Act lacks substantive mechanisms for protecting individual rights, fails to establish clear restrictions on AI applications posing risks of serious harm, and relies heavily on industry self-regulation in determining the content and effects of risk-based governance. Overall, the Act fails to perform the framework-setting function expected of a basic act and instead reads more like an industrial development white paper filled with policy declarations.

We solemnly urge lawmakers to confront the potential impact of AI technology on social equality and individual autonomy, rather than stopping at a development-policy document bearing the name of a "basic act," and to promptly launch the next stage of legislative reform to complete the unfinished work urgently required for AI governance:

I. Empowering the People in the AI Era: Promptly Codify the Rights to Information, Explanation of Decisions, and Human Review; Ensure the Rights to Equality, Remedy, and Participation in Policymaking

One important function of a basic act is to set out the rights that should be protected in the relevant field. Looking at Taiwan's legislative practice, the Fundamental Act of Education protects four rights, including the right to education,[1] while The Indigenous Peoples Basic Law protects rights including Indigenous peoples' access to broadcasting and media.[2] The draft Youth Basic Act, currently under review in the Legislative Yuan, further sets out five rights for young people.[3]

For an emerging field such as artificial intelligence, which still involves a high degree of uncertainty, the rights-protecting function of a basic act is especially important. Misuse of AI can push existing structural biases into self-reinforcing "feedback loops," automating injustice while leaving individuals without even the most basic tools for self-protection in the face of technological risks and power asymmetries. JRF therefore advocates that the AI Basic Act should expressly incorporate and protect the following six core rights:[4]

1. Right to Information: People have the right to know whether they are interacting with an AI system and to obtain information about whether that system complies with applicable laws, policies, and technical standards.[5]
2. Right to an Explanation of Decisions: When an AI-assisted or AI-generated decision has significant effects on an individual's rights or interests—such as in credit scoring, employment, or insurance—the individual should have the right to request a clear and understandable explanation sufficient to assess the decision's basis and reasonableness.[6]
3. Right to Human Review and to Opt Out of Fully Automated Decisions: Where significant rights or interests are affected, people should have the right not to be bound solely by an automated decision and to request meaningful human review.[7]
4. Right to Equality: AI systems must not reproduce or amplify discrimination based on gender, race, socioeconomic status, or other protected characteristics. Particular attention should be paid to preventing database bias from subjecting vulnerable groups, including Indigenous peoples, to "digital colonization" or intensified profiling and labeling.[8]
5. Right to Remedy: When AI development or use infringes people's rights, affected persons should have the right to seek cessation of the infringement and appropriate remedies, including civil damages or state compensation where applicable.[9]
6. Right to Participate in Policymaking: Communities affected by AI policies should be guaranteed meaningful opportunities for consultation and participation in the policymaking process.[10]

Regrettably, the Artificial Intelligence Basic Act passed by the Legislative Yuan contains no provision that expressly confers any of these rights on the people. JRF therefore calls on the Executive Yuan and lawmakers to begin drafting legal protections for three rights that are particularly important in the AI context—the right to information, the right to an explanation of decisions, and the right to human review—and to ensure through binding rules and concrete measures that the rights to equality, remedy, and participation in policymaking are fully realized in future AI governance.

At the same time, rights in the digital age should not be confined to the AI context but should extend across the broader digital environment. JRF is promoting the Digital Bill of Rights, a comprehensive effort to identify the rights people should enjoy and the obligations the state should bear in AI and the wider digital environment, with the goal of ensuring that the freedoms protected by the Constitution of the Republic of China and the requirements of democratic due process and the rule of law remain sustainable in the digital age.

II. Establishing a Safety Perimeter: Promptly Enact Binding Rules for AI Uses Posing Risks of Serious Harm

While actively promoting development and innovation, the Artificial Intelligence Basic Act, as an overarching basic law, should also draw clear and non-negotiable boundaries. Identifying categories of AI use that should be prohibited or strictly restricted is not only a fundamental safeguard for human rights; it also provides more concrete direction for subsequent legislation and gives industry a more predictable regulatory framework for investment and development.[11]

JRF believes that a sound basic act should at minimum declare that Taiwan's approach to prohibited or strictly restricted AI practices will be aligned with widely recognized international standards, so that the boundaries are neither excessively permissive nor unnecessarily restrictive, while balancing innovation with protection of the public.

JRF urges the Ministry of Digital Affairs, in light of its statutory role in developing Taiwan's AI risk-classification framework, to initiate further legislative work and propose concrete, legally binding governance rules for AI practices including the following:

1. Deploying AI systems that use manipulative or deceptive techniques, including systems that exploit vulnerabilities associated with age, physical or mental disability, or socioeconomic circumstances.
2. Establishing social-scoring systems.
3. Predicting an individual's risk of committing a criminal offense based solely or primarily on profiling or personality characteristics.
4. Building or expanding facial-recognition databases through indiscriminate scraping of facial images.
5. Using emotion-recognition systems in sensitive contexts such as education, employment, or criminal justice.
6. Using biometric categorization to infer sensitive characteristics, including race, political opinions, trade-union membership, religious or philosophical beliefs, sex life, sexual orientation, or gender identity.
7. Using biometric identification systems for law-enforcement purposes in public spaces.
8. Other AI applications that pose a significant risk to people's rights.

III. Rejecting the Law of the Jungle for AI: The Government Must Enact Substantive Rights-Protecting Legislation to Fill the Gaps Left by Industry Self-Regulation

The Artificial Intelligence Basic Act relies heavily on forms of "industry self-regulation" that lack independent legal force in its treatment of AI risks,[12] while bypassing or postponing issues that directly affect individuals and society as a whole. This creates a fragile and unbalanced foundation for Taiwan's vision of a democratic and free AI society.

The true test of AI governance has never been technology alone, but the value choices societies make about human beings. Debates over regulatory intensity ultimately concern relationships of power and competing interests: on one side are individuals whose information is collected and processed by AI systems and who may be subjected to automated decisions; on the other are the people and organizations that design, deploy, and profit from those systems.

Throughout the legislative process, misleading arguments repeatedly framed AI governance as a struggle between humans and machines, obscuring the fact that machines are designed and deployed by people—and often by enterprises with substantially greater economic resources and institutional power. The current predicament of the Artificial Intelligence Basic Act—having the form of a basic act while lacking sufficient substantive regulatory force—risks pushing Taiwan's AI environment toward a law of the jungle. In a regulatory environment without enforceable individual rights or clear boundaries that cannot be crossed, the central question becomes who is able to accumulate wealth and influence by exploiting the vulnerabilities, or even the growing dependency, of others.

We strongly urge the government, pursuant to Article 18 of the Artificial Intelligence Basic Act, to make "establishing digital rights through legislation" and "setting appropriate boundaries for prohibited or restricted AI applications" top priorities in the two-year process of regulatory review and adaptation, and to treat the construction of a human-centric AI legal framework as a principal legislative task during that period. Only by confronting these issues sooner rather than later—and by using democratic rule-of-law institutions to build the democratic rule of law of the future—can Taiwan ensure that AI development rests on a solid foundation of fairness, human-centric values, and protection of every person's rights.

---

Notes

1. Articles 7, 8, 14, and 15 of the Fundamental Act of Education.
2. Articles 12 and 20 of The Indigenous Peoples Basic Law.
3. Articles 6, 7, 8, 15, and 16 of the draft Youth Basic Act.
4. JRF's proposed provisions are as follows:Right to Information: "People shall have the right to know whether they are interacting with artificial intelligence and the right to obtain information about whether the artificial intelligence complies with applicable laws, policies, and technical standards."Right to an Explanation of Decisions: "Where a decision made by artificial intelligence produces legal effects or significant adverse effects on a person, that person shall have the right to request from the deployer an explanation of the individual decision."Right to Human Review and to Opt Out of Fully Automated Decisions: "Where a decision based solely on artificial intelligence produces legal effects or significant adverse effects on a person, that person shall have the right not to be bound solely by that decision and may request human review."Right to Equality: "People shall have the right to be free from discrimination arising from artificial intelligence decision-making processes or outcomes."Right to Remedy: "Where the research, development, or application of artificial intelligence infringes people's rights, affected persons may demand cessation of the infringement and shall have the right to seek civil damages or state compensation, as applicable."
5. See Article 50 of Regulation (EU) 2024/1689 (Artificial Intelligence Act); Section 1.3 of the OECD Recommendation of the Council on Artificial Intelligence; paragraphs 37 and 38 of UNESCO's Recommendation on the Ethics of Artificial Intelligence (2021); and Articles 8 and 15(2) of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law.
6. See Article 86 of Regulation (EU) 2024/1689 (Artificial Intelligence Act); paragraph 38 of UNESCO's Recommendation on the Ethics of Artificial Intelligence (2021); and Article 14(2)(b) of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law.
7. See Article 22 of Regulation (EU) 2016/679 (General Data Protection Regulation); Article 22 of the UK General Data Protection Regulation (UK GDPR); paragraph 38 of UNESCO's Recommendation on the Ethics of Artificial Intelligence (2021); and Article 15(1) of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law.
8. See Section 1.2 of the OECD Recommendation of the Council on Artificial Intelligence; paragraphs 19, 24, 31, 33, 38, 45, and 59(i) of the UN High Commissioner for Human Rights report The Right to Privacy in the Digital Age (A/HRC/48/31, 2021); paragraph 28 of UNESCO's Recommendation on the Ethics of Artificial Intelligence (2021); and Article 10 of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law.
9. See Article 99 of Regulation (EU) 2024/1689 (Artificial Intelligence Act); Article 82 of Regulation (EU) 2016/679 (General Data Protection Regulation); paragraphs 35 and 36 of UNESCO's Recommendation on the Ethics of Artificial Intelligence (2021); and Article 14(1)(c) of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law.
10. See Article 62(1)(d) of Regulation (EU) 2024/1689 (Artificial Intelligence Act); paragraphs 47 and 48 of UNESCO's Recommendation on the Ethics of Artificial Intelligence (2021); and Article 5(2) of the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law.
11. Article 16, Paragraph 2, latter part, of the Artificial Intelligence Basic Act: where circumstances listed in Article 5, Paragraph 1 exist, the relevant practices shall be restricted or prohibited in accordance with applicable laws and regulations. [Corrected on March 17, 2026]
12. Article 16, Paragraph 2, former part, of the Artificial Intelligence Basic Act: each competent authority for the relevant industry shall, according to the needs of AI application risk management and following the risk-classification framework referred to in the preceding paragraph, establish risk-based management rules and assist relevant industries in formulating industry guidelines and codes of conduct. [Corrected on March 17, 2026]

Contacts

Grace Huang, Attorney-at-Law / International Affairs and Digital Rights Project Manager, Judicial Reform Foundation

grace.huang@jrf.org.tw