Global Governance of Artificial Intelligence: Rule Competition, Regulatory Focus, and China's Response Strategy
Introduction
The rapid development of artificial intelligence has not only brought unprecedented productivity leaps to human society but also hidden deep risks such as algorithm runaway, technological alienation, and widening digital divides. In this context, major global economies are moving from conceptual discussions of "AI for Good" into substantive rule-building and strategic competition. Recently, China announced it is accelerating the establishment of the World AI Cooperation Organization (WACO), a multilateral institution to be based in Shanghai, aiming to jointly promote "AI for Good" and bridge the global AI divide. This move not only highlights China's active role in global AI governance but also reflects the international consensus on the urgency of AI regulation. This article delves into the landscape of global AI rule competition, regulatory focuses in specific governance dimensions, and proposes countermeasures for China to address global governance challenges.
Three Paradigms of Global Rule Competition
The leapfrog development of AI technology is profoundly reshaping global industries and political-economic landscapes. Beneath the surface of technological breakthroughs lie deep-seated institutional competition. Major economies, based on their own resource endowments, are accelerating the deployment of differentiated regulatory strategies, forming three dominant paradigms.
Europe: Defensive Normativism
The EU's regulatory path relies heavily on a comprehensive system of codified law. Its Artificial Intelligence Act (effective 2024) establishes a risk-based tiered regulatory model, classifying AI systems into unacceptable risk, high risk, limited risk, and minimal risk, with strict compliance reviews for high-risk applications such as recruitment, credit, and biometric identification. Its deep strategy lies in triggering the "Brussels Effect," attempting to convert its high internal privacy and ethical standards into global norms, exhibiting both defensive and normative characteristics.
United States: Pragmatism and Agile Pivot
The US regulatory path is more pragmatic and fragmented, with its core being maintaining absolute leadership in the global technology innovation chain. While the federal level continues an "innovation-first" tone, the Trump administration in June 2026 signed a new executive order establishing a "voluntary framework" requiring designated frontier models to provide the government up to 30 days of access before release, amid sharply rising risks of frontier models autonomously exploiting software vulnerabilities. This is seen as a safety pivot within a deregulation framework, reflecting a dynamic balance between risk and innovation.
China: Scenario-Based Governance Balancing Development and Security
China's regulatory path embodies the characteristic of balancing development and security. Targeting specific scenarios such as algorithmic recommendation, deep synthesis, and generative models, China has earlier issued dedicated regulations including the Internet Information Service Algorithmic Recommendation Management Provisions and the Interim Measures for the Management of Generative AI Services, implemented through a filing system. As of April 30, 2026, a total of 868 generative AI services have completed filing. This scenario-based agile governance model aims to precisely offset social risks while providing clear compliance expectations for domestic industries.
Regulatory Focus in Specific Governance Dimensions
The macro-level rule competition among major powers is cascading down to specific governance practices. International regulatory focus has shifted from abstract strategy frameworks to substantive application scenarios, with core attention concentrated on the following three key areas.
First: Establishing Technology Safety Baselines and Preventing Algorithm Runaway
Countries generally start from data privacy, algorithmic black boxes, and system robustness to prevent underlying technology runaway. Algorithmic bias has become a regulatory focus. For example, in the Derek Mobley v. Workday case, the plaintiff alleged AI recruitment tools systematically excluded older applicants. The court recently preliminarily certified the case as a class action and ruled that AI vendors cannot evade anti-discrimination liability on the grounds of "algorithmic decisions." This sends a clear signal: algorithmic bias is no longer a technical flaw but a legal red line directly subject to accountability.
Second: Upholding Human Autonomy and Preventing Technological Alienation
Global regulation is moving into deeper ethical waters, with the core being preventing "technological alienation." Article 14 of the EU AI Act requires high-risk automated decision-making systems to assign a natural person for effective oversight, with the right to question and correct system decisions. In May 2026, the EU postponed the enforcement deadline for core compliance obligations of high-risk systems to December 2027, providing flexible space for technology adaptation while holding the line on human autonomy. This "human-in-the-loop" principle clarifies that AI is ultimately an auxiliary tool, and final decision-making authority and accountability must rest with humans.
Third: Regulating Workplace Algorithmic Control and Protecting Worker Rights
As AI deeply embeds into task allocation, progress monitoring, and performance evaluation, workers face the risk of being fully controlled by algorithms. The EU Platform Work Directive, issued in December 2024, requires platforms to inform workers in writing of algorithmic decision-making logic, data collection scope, and impact on income, and prohibits processing sensitive data such as emotions and psychological states, with manual override required for key decisions. Meanwhile, New South Wales, Australia, passed the Digital Work Systems Act in February 2026, bringing AI, algorithms, and automated platforms under occupational health and safety regulation. These developments indicate that combating algorithmic alienation in the workplace and protecting workers' autonomy and control have become important issues in current governance practice.
China's Countermeasures and Recommendations for Addressing Global AI Governance Challenges
Facing increasingly complex international regulatory barriers and a widening global digital divide, a single regulatory model can no longer address systemic challenges. To better adapt to AI development trends, China needs to take pragmatic actions in three aspects while participating in global collaboration and optimizing domestic governance.
First: Actively Participate in International Technical Standards Development to Reduce Cross-Border Compliance Barriers
The dense and divergent legislation across countries objectively increases compliance costs for cross-border AI development. Given that universally binding international legal treaties are difficult to achieve in the short term, promoting mutual recognition of cross-border technical standards is a more pragmatic choice. Chinese industry organizations and technology companies can further strengthen engagement with professional technical platforms such as ISO and ITU, actively introducing and sharing domestic governance practices and technical standards in specific engineering aspects like data classification and grading, large model safety assessment, and interface compatibility.
Second: Implement Agile Governance Mechanisms like Regulatory Sandboxes to Balance Innovation and Safety
Over-reliance on fixed legal texts for ex-post regulation cannot fully adapt to the self-evolving technical characteristics of large models. China can further explore dynamic and adaptive agile governance, steadily expanding the coverage of innovation pilots like "regulatory sandboxes." Under clear safety baselines and controllable risk, provide moderate trial-and-error space for frontier model R&D. This companion-style regulatory interaction helps prevent systemic risks while better maintaining the domestic innovation ecosystem.
Third: Leverage the UN Framework to Promote an Inclusive and Equitable Global Governance System
The fairness and stability of global AI governance depend on broad international participation. China can continue supporting the UN's primary role in global digital governance, guided by the Global AI Governance Initiative, promoting the establishment of an inclusive multilateral coordination platform. In multilateral cooperation practice, systematically explore collaborative mechanisms covering joint construction of computing infrastructure, technology transfer, and high-quality data sharing, substantively responding to the real needs of developing countries to bridge the digital divide.
Conclusion
The global governance of artificial intelligence is both a test of countries' institutional wisdom and innovation capacity and a key to achieving sustainable development in the digital age. From rule competition to specific regulation, major economies are exploring different paths to balance innovation and safety. As an important participant and contributor, China needs to actively integrate into the international rule system while continuously optimizing domestic governance mechanisms, relying on multilateral platforms to promote a more inclusive and equitable global governance order. Only through consensus and coordinated action on a global scale can we truly achieve AI for Good, enabling AI technology to better serve the common interests of humanity.
