Policy Status
Policy status
In force
Effective date
January 2026
Applies to
Both
Regulatory impact
Medium
north-asia
Taiwan
binding law
Quick Overview
- Taiwan's Legislative Yuan passed the Artificial Intelligence Basic Act (人工智慧基本法) on 23 December 2025, making it one of Asia's first comprehensive standalone AI laws.
- The Act was promulgated by President Lai Ching-te on 14 January 2026 and entered into force immediately, establishing a principles-based governance framework grounded in seven core AI ethics standards.
- The National Science and Technology Council (NSTC) serves as the central competent authority, while the Ministry of Digital Affairs (MODA) is tasked with developing an internationally aligned risk classification framework.
- Taiwan's approach deliberately balances innovation promotion with human-centric safeguards, positioning the island as a potential regulatory model for other economies in the Asia-Pacific region.
What's Changing
- The AI Basic Act codifies seven core principles into law: sustainability and well-being, human autonomy, privacy and data governance, cybersecurity and safety, transparency and explainability, fairness and non-discrimination, and accountability.
- A National AI Strategy Special Committee under the Executive Yuan will coordinate cross-ministerial AI policy, overseeing both innovation promotion and risk governance.
- MODA is developing a risk-based classification framework — expected to be released in Q1 2026 — that will determine which AI applications qualify as high-risk and require enhanced oversight.
- The government has allocated NT$30 billion for 2026 AI initiatives as part of a broader NT$100+ billion multiyear investment programme covering research, talent development, and digital infrastructure.
- AI-generated deepfakes and synthetic media will face labelling requirements, with sector-specific regulations expected to follow in healthcare, finance, and public services.
- Regulatory sandboxes are being established to allow controlled testing of innovative AI applications before full market deployment.
Who's Affected
- AI developers and service providers operating in Taiwan must align with the Act's seven core principles, with more detailed compliance requirements expected once MODA's risk classification framework is finalised.
- Government agencies at all levels are required to integrate AI ethics standards into public service delivery, procurement, and decision-making processes.
- Taiwan's semiconductor and technology companies — including firms in the TSMC supply chain — will benefit from increased R&D funding and regulatory clarity for AI chip development.
- Healthcare providers, financial institutions, and educational platforms deploying AI systems may face enhanced oversight obligations once high-risk classification criteria are defined.
- Foreign technology companies offering AI services in Taiwan should monitor the evolving regulatory framework to ensure compliance as implementation details emerge.
- Citizens gain principle-based protections against unfair AI-driven decisions, with specific redress mechanisms expected to be detailed in subsequent regulations.
Core Principles
- Sustainability and well-being: AI development must contribute to sustainable social and economic outcomes, with consideration for environmental impact and intergenerational welfare.
- Human autonomy: Individuals retain the right to meaningful human control over AI-assisted decisions that significantly affect their lives, with AI serving to augment rather than replace human judgement.
- Privacy and data governance: AI systems must respect personal data protection standards, with clear rules on data collection, processing, and retention for AI training and operation.
- Cybersecurity and safety: AI systems must meet robust security standards to prevent misuse, unauthorised access, and system failures that could endanger public safety.
- Transparency and explainability: AI operators must provide meaningful information about how AI systems function, particularly for decisions affecting individual rights and opportunities.
- Fairness and non-discrimination: AI systems must be designed to prevent algorithmic bias and must not discriminate based on protected characteristics.
- Accountability: Clear lines of responsibility are established for AI outcomes, ensuring that individuals and organisations can be held accountable for harm caused by AI systems.
What It Means for Business
- Companies operating in Taiwan should begin aligning their AI governance practices with the Act's seven principles now, even before detailed implementation regulations are issued.
- The risk classification framework from MODA (expected Q1 2026) will be critical — businesses should prepare for the possibility that their AI applications may be classified as high-risk, particularly in healthcare, finance, and hiring.
- Taiwan's substantial AI investment programme (NT$30 billion in 2026 alone) creates significant opportunities for AI startups, research collaborations, and government-backed innovation projects.
- Regulatory sandbox provisions allow businesses to test novel AI applications under supervised conditions, reducing risk and accelerating time-to-market for qualifying products.
- The Act's principles-based approach provides flexibility for businesses to develop compliant practices without overly prescriptive rules, though this may change as sector-specific regulations emerge.
- Given Taiwan's critical role in global semiconductor supply chains, compliance with the AI Basic Act positions companies well for international partnerships and cross-border AI governance alignment.
What to Watch Next
- MODA's risk classification framework is the most anticipated next step — expected in Q1 2026, it will define which AI applications are considered high-risk and what enhanced compliance obligations apply.
- Sector-specific implementation guidelines for healthcare AI, financial services automation, and public sector AI deployment are expected throughout 2026.
- The National AI Strategy Special Committee's first policy recommendations will shape how Taiwan balances its innovation ambitions with regulatory safeguards in practice.
- International alignment discussions — particularly with the EU AI Act, Japan's AI governance framework, and South Korea's AI Basic Act — will influence how Taiwan calibrates its regulatory approach for cross-border interoperability.
- Talent development initiatives under the NT$100+ billion investment programme will create new opportunities for AI researchers, engineers, and governance professionals.
- Ongoing legislative discussions about amendments to the Personal Data Protection Act may introduce additional requirements for AI training data governance and cross-border data transfers.
← Scroll to see full table →
| Aspect | Taiwan | Japan | South Korea |
|---|---|---|---|
| Approach Type | Draft legislation | Principles | Rights-based |
| Legal Strength | Pending | Voluntary | Moderate |
| Focus Areas | Accountability, risk levels | Fairness, safety | Privacy, fairness |
| Lead Bodies | National Development Council | METI, Cabinet Office | MSIT, PIPC |
Local Resources
Last editorial review: March 2026
Related coverage on AIinASIA explores how these policies affect businesses, platforms, and adoption across the region. View AI regulation coverage
This overview is provided for general informational purposes only and does not constitute legal advice. Regulatory frameworks may evolve, and readers should consult official government sources or legal counsel where appropriate.
