Taiwan’s AI Strategy and Regulatory Framework
The government of Taiwan has adopted a proactive approach to support the AI industry, promoting industrial development through policy measures and corresponding legal frameworks. In the latter half of 2024, the National Science and Technology Council (NSTC) introduced the draft AI Basic Act, which was submitted to the Executive Yuan (Taiwan’s cabinet) for review in early 2025.
In parallel, Taiwan has amended laws to address AI-driven fraud, deepfake activities, and election manipulation. The government also plans to enact new legislation on data governance and open data to address the data-driven characteristics of AI.
AI Government Policies
Taiwan’s government actively supports the development of professional AI chips, AI hardware, and large-scale language models to promote the comprehensive growth of AI research and applications. Meanwhile, sectors such as manufacturing, finance, healthcare, agriculture, and retail are encouraged to integrate AI for digital transformation.
Using the TAIWANIA 2 supercomputer, the National Applied Research Laboratories launched TAIDE, a large-scale localized language model tailored to Taiwanese data. TAIDE utilizes public data (including judgments, Constitutional Court interpretations, and other court decisions from Taiwan’s Judicial Yuan) to refine traditional Chinese-language models. The model supports languages such as Taiwanese and Hakka, aiming to integrate AI into the agriculture, education, and automation industries.
This center will establish certification mechanisms and guidelines for AI products, as well as systems to ensure safer and more interpretable AI applications.
Legal Responses to AI Challenges
Despite rapid advances in AI technology, legal challenges remain. The Legislative Yuan is prioritizing cases where AI or deepfake technology is used for fraudulent or election manipulation purposes. Meanwhile, the Ministry of Digital Affairs (MODA) is drafting and revising legal frameworks for data governance. The NSTC’s draft AI Basic Act is intended to lay the groundwork for interagency collaboration and unified regulation of AI. These efforts fall into three core areas:
- Recognizing AI Risks: Relevant amendments have been made to the Criminal Code, the Fraud Crime Hazard Prevention Act, and other laws to establish criminal liability for disseminating false information or committing crimes using deepfake technology. Online advertising platforms are required to disclose instances where such technology is employed.
- New Legislation on Data Innovation: MODA is drafting the Act for the Promotion of Data Innovation and Utilisation, aimed at increasing the accessibility of open data and establishing cross-industry data-sharing mechanisms.
- Mitigating Data Use Risks: The use of data is subject to the Copyright Act and the Personal Data Protection Act. The Intellectual Property Office has ruled that using AI technology to generate output without the copyright holder’s consent may constitute reproduction of others’ works.
In parallel, the draft amendments to the Personal Data Protection Act will be rolled out with substantially updated provisions, enhancing personal data protection in an era where “data is the new oil of the digital economy.”
Draft Fundamental Act on Artificial Intelligence
To ensure that AI technology aligns with human rights, privacy, industrial competitiveness, and the public interest, the NSTC introduced the draft AI Basic Act in 2024, which is expected to be enacted in 2025. Key elements of the draft include:
- Definition and Scope of AI: The definition of AI is crucial as it determines the scope of regulation. The draft ensures broad coverage of AI techniques, from basic knowledge-based algorithms to sophisticated neural networks.
- Guiding Principles of AI: The draft sets out guiding principles for AI R&D, including sustainability, human autonomy, privacy, data governance, security, transparency, explainability, fairness, and accountability.
- Risk-Based Management: MODA will classify AI risks in line with international standards, promoting AI innovation within safety parameters.
- Data Privacy and Openness: Data openness and governance must be mandated to ensure the availability of adequate data for AI models while protecting personal data.
- Adaptive Legislation and Cross-Agency Collaboration: Each ministry must review its regulatory framework to ensure alignment with the rapid technological evolution of AI.
Conclusion
Taiwan plays a critical role in the global AI landscape with its advanced ICT and semiconductor industries, demonstrating the government’s effective policymaking from nurturing AI talent to refining AI laws. Taiwan is diligently building a comprehensive policy and legal framework around AI, showcasing its commitment as it advances into the AI era.
