While the artificial intelligence race is often described as a bipolar competition between Beijing and Washington, Taiwan, Japan, and South Korea are each seeking to secure a leading position of their own.
Within the span of a few months, Japan rallied a consortium of domestic tech giants behind a state-backed sovereign AI company, South Korea unveiled a US$576 billion investment plan to build out chip manufacturing and artificial intelligence capabilities, and Taiwan launched its New Ten Major AI Infrastructure Projects to become, in the words of President Lai Ching-te (賴清德), an “AI island.”
Each of the three governments is racing to establish itself as Asia’s leading AI hub. In that race, however, the guardrails that should frame their innovation goals are being left behind.
The pattern is visible in how each has legislated. Japan’s AI Basic Plan, adopted in December 2025, frames the national priority as deploying AI quickly enough to matter rather than constraining its development.
Taiwan’s Artificial Intelligence Basic Act, promulgated in January 2026, establishes guiding principles while imposing few hard legal obligations.
South Korea’s Basic Act on the Development of Artificial Intelligence and the Establishment of a Foundation for Trustworthiness, in force since January 2026, stands as the region’s only binding framework. Even there, the law grants at least a one-year grace period before penalties apply and was calibrated with evident care not to burden domestic industry.
The result is a collective action problem. Three advanced democracies, each capable of building robust AI governance on its own, have rational reasons to keep their guardrails light to foster innovation, with each aiming to become the third major actor in AI.
Cooperation offers a way out of this dilemma. A light coordination framework could make basic safeguards less costly to adopt while leaving regulatory authority in national hands.
Learning From Chip 4’s Limits
Early steps should remain light to account for divergent policy priorities, political tensions, and diplomatic challenges.
Any such initiative should begin by learning from the experience of the Chip 4 alliance. Proposed in 2022, the framework brought the US, Japan, South Korea, and Taiwan together around semiconductor supply chain resilience.
However, Chip 4 has not developed into the substantive and institutionalized framework its name appeared to promise, with little policy coordination since its establishment because of political tensions among members and pressure from China.
AI cooperation could suffer the same fate if these past difficulties are not addressed.
Taiwan’s diplomatic status creates a further difficulty. Japan and South Korea lack formal diplomatic relations with Taipei, and a formal governmental trilateral body would invite coercion from Beijing.
Nonetheless, the potential for Taiwan, Japan, and South Korea to cooperate on AI governance and represent a concrete third way in the global AI ecosystem is substantial if sufficient political will exists.
The three are technological pillars of AI hardware: Taiwan as a semiconductor powerhouse, Japan as a central actor in semiconductor manufacturing equipment, and South Korea as a leading producer of high-bandwidth memory.
Japan and Taiwan have already shown that stronger collaboration over key technological assets is achievable. Despite diplomatic constraints, their cooperation on semiconductor investment and supply chain resilience has deepened.
Beyond complementary technological strengths, Taiwan, Japan, and South Korea also share democratic values, a commitment to human rights, and common regional security concerns.
Industry leaders and researchers have also called for deeper collaboration among the three sides to move beyond the binary choice between China and the US.
The incentives for building a cooperative framework for AI governance are strong. What is needed now is concrete action.
Regular exchanges among regulators, shared research on emerging risks, and common methods for evaluating advanced AI systems are first steps that could give each government greater confidence that stronger safeguards would be matched by comparable expectations across the region.
Building an Early AI Alliance
A practical plan could proceed in three stages.
First, through the semi-official channels already used in relations with Taiwan, the three sides should establish an annual working-level AI governance dialogue that rotates among Taipei, Tokyo, and Seoul and is supported by industry groups, academia, technical agencies, and research centers.
Second, technical working groups should pursue a small set of shared outputs. These could include common standards and norms for advanced models and a shared mechanism for reporting serious AI incidents across borders.
Third, bilateral projects should provide the building blocks for wider coordination. Japan and Taiwan could extend their technology partnership into AI evaluation and governance.
Japan and South Korea could deepen their existing discussions on trustworthy AI, while exchanges between Taiwan and South Korea could connect the two sides through Track 1.5 and Track 2 diplomacy. These channels could gradually converge as trust develops.
Such steps could establish common expectations for transparency, independent testing, and incident reporting for frontier models while building trust among the three democracies.
Chip 4 demonstrates the limits of cooperation built around an expansive geopolitical label. Japan and Taiwan demonstrate what focused cooperation can achieve despite diplomatic constraints.
The next step is to apply that lesson to AI before the region’s governance gap becomes as consequential as its technological ambitions.
