In July, Taiwan’s National Science and Technology Council (NSTC) unveiled a draft proposal for an AI law, representing a significant step in the government’s commitment to embracing the potential of artificial intelligence while ensuring its development is aligned with global standards and ethical considerations.
Central to the basic AI act is the balance between promoting AI development and managing its risks. This approach reflects Taiwan’s intention to establish itself as a leader in AI innovation without stifling creativity or overburdening companies with excessive regulations. AmCham Taiwan commends the government for its comprehensive approach, particularly its emphasis on public-private partnerships and international collaboration, both of which are crucial for staying aligned with the global AI landscape.
However, concerns have been raised about the act’s vagueness and lack of specificity in some areas, potentially hindering its effectiveness. One of the most significant criticisms is the act’s ambiguous definition of AI. Echoing concerns raised during the development of the EU AI Act, Taiwan’s draft legislation fails to clearly define what constitutes artificial intelligence, creating confusion among companies over which regulations will apply to them. This obscurity is particularly problematic for those on the “user end,” who may not be directly involved in developing AI but use it in their operations.
In its current form, the act focuses more heavily on regulating AI development than its deployment and use. While it is critical to hold AI developers accountable, the failure to address user-end responsibilities creates a gap in ensuring trustworthy AI across the entire AI lifecycle – from research and development to implementation and use. This gap raises questions about the act’s ability to foster responsible AI usage across sectors effectively.
Further, the legislation needs to provide more guidance on which types of algorithms should be regulated, leaving companies uncertain about compliance requirements. This ambiguity could stifle innovation, as companies may hesitate to invest in AI research and development due to concerns over compliance risks and potential penalties.
Another concern raised by stakeholders is the act’s piecemeal nature. Critics argue that the draft lacks a coherent strategy tailored to Taiwan’s unique needs and circumstances. Rather than focusing on the specific challenges and opportunities within Taiwan’s AI ecosystem, the act appears to adopt a one-size-fits-all approach that may not be suitable for Taiwan’s distinct technological and economic context.
AmCham committees have proposed several recommendations to address these shortcomings. A key suggestion is to hold public hearings before the act is submitted to the Executive Yuan. Such hearings would enable broader input from diverse stakeholders, ensuring the legislation is informed by the latest technological developments and industry insights.
Another important recommendation is to adopt a principle-based, risk-based regulatory framework. Instead of imposing prescriptive rules, the AI Basic Act should offer flexible, high-level guidelines that can be adapted as AI technology evolves. Such an approach would help prevent overregulation while allowing companies the freedom to innovate, all while ensuring accountability for high-risk AI applications.
Additionally, to avoid inconsistencies across government agencies and promote a more coherent regulatory approach, AmCham recommends that NSTC be designated as the main coordinating body for AI regulation.
By incorporating stakeholder feedback, adopting a flexible and risk-based approach, and clearly defining AI and the responsibilities of developers and users, Taiwan can ensure that its new AI act fosters innovation and trust.
