Taiwan Artificial Intelligence Basic Act

Passed 23 December 2025 and in force 14 January 2026. Codifies seven principles (sustainability, human autonomy, privacy, security, transparency, fairness, accountability), mandates risk-tier labelling of high-risk AI applications, data openness with privacy by design, and labour safeguards.

JurisdictionTaiwan
CategoryAI Regulations
StatusActive
Effective date
Latest development

Analysis

Taiwan’s Artificial Intelligence Basic Act is a top‑level framework statute that sets principles and mandates government action on AI governance rather than a detailed, directly enforceable rulebook for private companies.Artificial Intelligence Basic Act – NSTC A basic law for AI – Executive Yuan AI Watch – Taiwan, White & Case

Below is a structured, compliance‑oriented analysis in the format you requested. Because the official English text is relatively concise and many operational details are being developed by ministries, some practical interpretations draw on expert analyses, which are explicitly linked.

Key Requirements

1. Seven AI Governance Principles

2. Effective Date and Scope

3. Risk‑Tier Classification and Labelling of High‑Risk AI

4. Data Openness with Privacy by Design/Default

5. Labour and Social Safeguards

6. Institutional Architecture and Government Duties

7. Timelines for Government Implementation

Compliance Challenges

Because the Act is largely framework‑oriented, compliance challenges relate both to interpretation and to anticipating future sectoral rules.

1. Uncertainty for Private‑Sector Obligations

2. Risk‑Tier and High‑Risk AI Identification

3. Integrating Principles into Existing Governance

4. Examples of Emerging Practice and Difficulties

Implementation Best Practices

These practices are not legally binding but align with the Act’s principles and the direction of expert guidance.

1. Establish an AI Governance Framework Aligned to the Seven Principles

  • Organizations can adopt a formal AI governance framework mapping each internal policy and control to the seven principles (sustainability, autonomy, privacy, security, transparency, fairness, accountability).[AI Basic Act – MODA Governance Page](https://moda.gov.tw/en/major-policies/ai/governance/192

Recent developments

  • — A regulatory tracker notes that the **Artificial Intelligence Basic Act** has been in force since 2026‑01‑14 and that government agencies must complete risk assessments for all public‑facing AI applications by July 2026, signaling concrete implementation deadlines and internal‑control obligations for the public sector.[8] (source)
  • — Taiwan’s Ministry of Digital Affairs outlines governance and evaluation under the **Artificial Intelligence Basic Act**, emphasizing seven core principles (sustainable development, human autonomy, privacy/data governance, cybersecurity, transparency, fairness, accountability) and policy incentives to drive secure AI innovation in the public sector.[6] (source)
  • — A legal analysis describes the **AI Basic Act** as Taiwan’s 2026 roadmap for AI, highlighting its seven guiding principles and a two‑year program for reviewing and amending existing laws and administrative measures to align with the Act, setting the foundation for future sector‑specific regulation and compliance expectations.[4] (source)
  • — The National Science and Technology Council publishes the English text of the **Artificial Intelligence Basic Act**, a 20‑article framework law effective from 2026‑01‑14 that establishes human‑centric AI development, digital equity, rights protection, social well‑being, and the obligation for government to ensure AI use complies with social ethics.[1] (source)
  • — The Executive Yuan announces the **National Artificial Intelligence Strategy Special Committee** to oversee AI affairs and implement the AI Basic Act, requiring all government agencies to complete AI‑use risk assessments by July 2026 and to adopt internal control rules within one year, accelerating compliance work across the public sector.[2] (source)
  • — A governance analysis describes the AI Basic Act as a 20‑article framework that sets seven principles, creates a multi‑agency coordination architecture, and gives sector regulators two years to develop detailed rules, positioning Taiwan as a potential model for **democratic AI governance** in Asia while still leaving many implementation questions open.[12] (source)
  • — An opinion piece argues that Taiwan’s **AI Basic Act** provides important guardrails around transparency, accountability, and safety but remains weak because it is largely non‑binding for the private sector, prompting calls from experts for stronger obligations and clearer enforcement mechanisms in future implementing rules.[3] (source)
  • — A policy commentary argues that Taiwan’s **AI Basic Act** can serve as a model for Asia by combining innovation‑friendly principles with risk‑based governance and multi‑stakeholder coordination, while noting that its ultimate impact on the AI industry will depend on how forthcoming implementing measures address gaps such as enforcement and private‑sector duties.[13] (source)
  • — A global AI regulatory tracker notes that Taiwan’s **AI Basic Act** entered into force on 2026‑01‑14 as the legal basis for regulating and promoting AI, but stresses that it imposes no direct obligations on private‑sector companies, meaning near‑term impacts fall mainly on government agencies while industry awaits sector‑specific regulations.[11] (source)
  • — A legal briefing explains that the AI Basic Act designates the NSTC as the central authority, mandates creation of a National AI Strategy Committee, and tasks MODA with an internationally aligned risk‑classification framework, setting timelines (3, 6, 12, and 24 months) for impact assessments, government AI risk reviews, internal control rules, and legislative updates that will shape future compliance and market practices.[7] (source)

Related regulations

Put it into practice

Browse all regulations · Compliance deadlines · Latest updates