Key takeaway: Taiwan's first dedicated AI law, the Artificial Intelligence Basic Act (AI Basic Act), passed its third reading in the Legislative Yuan on December 23, 2025 and was promulgated by the President on January 13, 2026, with 20 articles in total. The central competent authority is the National Science and Technology Council (NSTC); the local competent authorities are the special municipality and county (city) governments. As a "basic act," it declares seven fundamental principles and calls for an internationally aligned AI risk classification framework; the detailed rules that will directly affect businesses will be introduced progressively by the sector-specific competent authorities under that framework.
1. Legislative History and Positioning: Taiwan's First Dedicated AI Law
On December 23, 2025, the Legislative Yuan passed the AI Basic Act on third reading — Taiwan's first dedicated artificial intelligence law, comprising 20 articles; the President promulgated it on January 13, 2026. The bill had earlier been approved by the Executive Yuan and sent to the Legislative Yuan for deliberation before completing the legislative process. As for competent authorities, at the central level it is the National Science and Technology Council (NSTC); at the local level, the special municipality and county (city) governments.
The key to understanding this law lies in the words "basic act." It is not a regulatory statute that directly imposes specific obligations or penalties on businesses; it is the top-level guiding framework for national AI policy — declaring principles, setting direction, and serving as the basis for subsequent legislation and policy by individual ministries. In other words, the detailed rules that will actually shape day-to-day operations in each industry will be introduced progressively by the sector-specific competent authorities under the risk framework. The passage of the Basic Act is the starting point of Taiwan's AI legal architecture, not the end point.
Notably, the Act also brings labor rights into scope: the government must use AI in ways that protect workers' rights and interests, close skills gaps, raise labor participation, safeguard economic security and realize decent work; for workers who lose their jobs due to the use of AI, the government must provide employment assistance in line with their ability to work. This reflects the legislature's pre-emptive response to AI's impact on the job market — businesses introducing AI as part of organizational change would do well to factor these policy directions into their planning.
2. The Seven Fundamental Principles: A Common Language for AI Governance
The Act declares seven fundamental principles the government must follow in promoting AI research, development and application. Although these principles are addressed primarily to the government, they can be expected to become the common baseline as competent authorities write the rules for each industry — and they are the best reference points for building your company's internal AI governance:
| Fundamental principle | Meaning and the business perspective |
|---|---|
| Sustainable development and well-being | The development and application of AI should contribute to the sustainable development of society as a whole and to human well-being, not merely pursue technical or commercial gains. |
| Human autonomy | AI should assist rather than replace autonomous human decision-making; when designing AI products and services, businesses should preserve appropriate room for human judgment and oversight. |
| Privacy protection and data governance | Data collection and use for AI must respect privacy; this connects directly with the personal data protection regime and is a top compliance priority for businesses. |
| Cybersecurity and safety | AI systems must ensure information and communication security and safety in use, guarding against misuse, attacks and loss-of-control risks. |
| Transparency and explainability | How AI operates and what it produces should be understandable and explainable; businesses must be able to account for the basis of AI decisions to customers and regulators. |
| Fairness and non-discrimination | AI applications should avoid discrimination caused by algorithmic bias — recruitment, credit and pricing scenarios deserve particular care. |
| Accountability | Clear responsibility must be established for AI conduct and outcomes; internal governance should designate responsible units and accountability mechanisms. |
※ The authoritative meaning of each principle is as set out in the AI Basic Act and interpreted by the competent authorities; the business-perspective notes in the table are our firm's practical observations.
3. The Risk Classification Framework: Article 16 and the Role of the Ministry of Digital Affairs
For businesses, the provision most worth tracking is Article 16: the Ministry of Digital Affairs must, with reference to international standards, promote an AI risk classification framework aligned with international practice, and assist the sector-specific competent authorities in formulating risk-based management rules.
This means Taiwan's future AI regulation will follow a risk-tiered approach — AI applications at different risk levels will face management requirements of different intensity: low-risk applications will not be over-regulated, while high-risk applications will bear stricter obligations. And because the framework emphasizes international alignment, the compliance investments businesses make now are more likely to carry over to international market requirements, avoiding duplicated effort.
In terms of division of labor, the Ministry of Digital Affairs drives the risk classification framework, while the concrete management rules for each industry are written by that industry's own competent authority under the framework. The implication: businesses in different industries may face different AI rules on different timelines — tracking your own industry regulator matters more than tracking the Basic Act itself.
4. Five Practical Implications for Businesses
First, map your AI applications and risk levels: build an internal inventory of AI use — which teams use which AI, for which decisions, affecting whom — and make a preliminary risk assessment of each application as the basis for what follows. Second, watch your industry regulator's forthcoming rules: detailed regulation will come progressively from the sector-specific competent authorities, and highly regulated industries such as finance and healthcare should monitor especially closely. Third, build internal AI governance: matching the principles of transparency and explainability, fairness and non-discrimination, and accountability — adopt internal use rules, designate responsible units, and establish review and record-keeping mechanisms. Fourth, connect personal data and cybersecurity compliance: privacy protection and data governance, and cybersecurity and safety, are already among the seven principles; AI governance should not be decoupled from your existing personal data and cybersecurity compliance — review them together. Fifth, prepare early to lower adjustment costs: while the detailed rules are still taking shape, get governance documents, training and vendor management in place step by step, so you are not scrambling when the rules land.
5. How LegalAI Helps Businesses Navigate AI Regulation
LegalAI Law Firm is itself a law firm that applies AI deeply in its own practice, giving us a front-line understanding of where AI technology meets the law. We provide AI governance and cybersecurity compliance advisory services: helping businesses map AI application risks, build internal AI governance and use rules, connect personal data and cybersecurity compliance, and continuously track forthcoming rules from the competent authorities — keeping your business ahead of compliance as Taiwan's AI legal framework takes shape.
Frequently Asked Questions
When was the AI Basic Act passed and promulgated?
The AI Basic Act passed its third reading in the Legislative Yuan on December 23, 2025, becoming Taiwan's first dedicated AI law, with 20 articles in total; the President promulgated it on January 13, 2026. The central competent authority is the National Science and Technology Council (NSTC), and the local competent authorities are the special municipality and county (city) governments.
Will the AI Basic Act directly penalize businesses?
In principle, no. As a basic act, the law is positioned as a guiding framework for subsequent legislation and policy by individual ministries, declaring principles and policy directions rather than directly imposing specific obligations or penalties on businesses. Detailed regulation of specific AI applications will be introduced progressively by the sector-specific competent authorities under the risk classification framework, so businesses should watch for forthcoming rules from the regulator of their own industry.
What is AI risk classification?
Under Article 16 of the AI Basic Act, the Ministry of Digital Affairs must, with reference to international standards, promote an AI risk classification framework aligned with international practice and assist the sector-specific competent authorities in formulating risk-based management rules. In short, AI applications are classified by level of risk — the higher the risk, the stricter the management requirements an application will face in the future.
What should businesses do now to prepare?
Businesses are advised to start by mapping their internal AI use cases and likely risk levels, keep monitoring forthcoming rules from the competent authority of their own industry, and begin building internal AI governance systems that implement principles such as transparency, explainability and accountability, while reviewing personal data protection and information security compliance alongside them to reduce future adjustment costs.
References
- Central News Agency (December 23, 2025): Legislative Yuan passes AI Basic Act on third reading, with the NSTC as competent authority
- Ministry of Digital Affairs press release: Legislative Yuan passes the AI Basic Act on third reading
- Executive Yuan: Cabinet approves the draft AI Basic Act
This article was compiled by our lawyers with the assistance of LegalAI's AI legal assistants, based on the public sources listed above. It is for general reference only and does not constitute legal advice on any specific matter. For the application of specific obligations, please rely on the competent authorities' announcements and the official text of the laws, or consult our lawyers.