China's AI Governance: A Unified AI Law No Longer in Sight?

Third Plenum Reaffirms the Need for AI Regulation

Until recently, it appeared that China, unlike the US, was poised to follow the EU's lead by passing a unified AI law, as was already indicated in the 2017 New Generation AI Development Plan. An early draft of a national AI law was submitted for review to the National People’s Congress (NPC) Standing Committee in June 2023. Now, it seems that China may delay establishing or abandoning a unified AI law. According to the South China Morning Post, the primary reasons include significant disagreement among government, industry, and academic communities on the regulatory approach, the rapid evolution of AI technology making comprehensive legislation premature, and a strategic preference for sector-specific regulations that may offer more flexibility and adaptability. The ongoing debate over intellectual property rights concerning data used or generated by AI systems has not led to a consensus either. For now, this would leave the EU as the only major economic bloc with a comprehensive AI law, which has received mixed feedback, with some arguing that it will stifle the EU's already limited competitiveness in AI.

Despite the absence of a unified law, China has still followed an issue-based approach, passing several regulations on AI safety. Notably, the "Interim Measures for the Management of Generative Artificial Intelligence Services" in 2023 require generative AI providers to register with authorities, manage content, protect data, ensure algorithm transparency, adhere to ethical standards, implement safety measures, comply with regulatory enforcement, and align with Chinese socialist values, avoiding politically sensitive topics. While a unified AI law has been anticipated towards the end of 2024, it may still be too premature to conclude that China is altogether abandoning its plan to establish such legislation.

Some indication is provided by The Third Plenary Session, which was held from July 15 to 18, 2024, and serves as a recurring mechanism for strategic governance and reforms. This Third Plenum of the 20th Central Committee of the Communist Party of China addressed all critical areas of reform and modernization, including AI innovation and governance. The Communique, Resolution, and President Xi’s Explanation do not explicitly mention the immediate adoption of a unified AI law. However, they still imply a move towards a stronger regulatory framework and oversight mechanism for AI, confirming the existing approach and suggesting the potential development of more comprehensive AI-specific regulations or laws in the future. In a balanced approach, the Third Plenum also emphasized innovation and supporting an environment that is conducive to technological breakthroughs and self-sufficiency. Key points from the session include (the first four focus on promoting innovation, the latter two on safety and control):

1. Supporting All-Around Innovation:

   1. Scientific and Technological Structural Reform: Promote innovation in key generic technologies, including AI.

2. Promoting High-Quality Economic Development:

   1. Facilitating Revolutionary Technological Breakthroughs: Boost policy and governance systems for strategic industries such as next-generation IT and AI, aviation and aerospace, new energy, new materials, high-end equipment, biomedicine, and quantum technology.

3. Pursuing High-Standard Opening Up:

   1. Improving Mechanisms for High-Quality Cooperation: Enhance collaboration under the Belt and Road Initiative and develop multilateral platforms for cooperation in areas including AI.

4. Improving the System for Comprehensive Cyberspace Governance:

   1. Developing Mechanisms for Managing Generative AI: Improve long-term governance mechanisms for the online environment, including AI applications.

5. Balancing Development and Security Imperatives:

   1. Instituting Oversight Systems: Ensure the safety of AI through robust oversight mechanisms.

This indicates that the idea of a unified AI law has not been fully abandoned but still requires time to develop. The regulatory uncertainty is also a result of the fierce competition and external pressures. China has been facing tightening restrictions on access to U.S. advanced technology, including semiconductors and AI, amid a rivalry marked by American claims of security risks and unfair subsidies. While the U.S.’ containment strategy hurts China, a recent CSIS study casts doubt on its effectiveness in containing China’s development. Accordingly, U.S. semiconductor export controls are harming the domestic industry by reducing revenue, market capitalization, and competitiveness of U.S. firms, while accelerating China's innovation and self-sufficiency in semiconductor technology, particularly in mature chips, where China's market share is poised to rise from an already significant 31%. The chairman of the China Semiconductor Industry Association has forecasted "explosive growth" in the next three to five years, leveraging the country's strengths in applications and packaging technologies. This growth trajectory is thought to help China overcome U.S. technology restrictions.

The trade and geopolitical tensions between China and the U.S. complicate reaching consensus on a unified AI law, as such a law could compromise China’s own competitiveness and race towards technology independence. The Chinese government identifies AI technology as one of the “new quality productive forces” crucial for future growth and development. Despite these grave tensions, China, the U.S., and the EU still have something in common. They all share to some extent a risk-based approach towards AI governance, yet with significant differences. Arguably, China’s approach lies between that of the U.S. (market-friendly, national security-focused) and the EU (strict regulatory, fundamental rights-focused), adopting a centralized strategy while balancing innovation with social stability and national security. As a unified law remains elusive, China is closer to the U.S. model, which lacks a national AI law but also follows an issue based approach still manifested in President Biden’s Executive Order on AI. While the Executive Order risks being revoked by a newly reelected Trump administration, over 70% of Democrats and Republicans support many of the initiatives in the executive order. The Republican New Right movement might find unexpected common ground with China in their shared belief that Big Tech and its technologies should serve the real economy and society rather than being driven by capital markets. This perspective aligns with views expressed during the Third Plenum, which emphasized the need for technology to support broader societal and economic goals over capital market-driven priorities. The Chinese government's crackdown on Big Tech, which began in late 2020, was driven by concerns that major internet platforms had become too powerful and posed risks to financial stability, data security, and fair market competition.

The Shanghai Declaration on Global AI Governance, announced on July 4, 2024, during the World Artificial Intelligence Conference (WAIC), underscores the importance of national sovereignty and priorities in AI safety. The Shanghai Declaration primarily pledges global cooperation on AI governance to ensure the safe, reliable, and fair development and application of AI technologies. However, the absence of national AI laws, with the exception of the EU, complicates governance interoperability and the establishment of an enforceable global AI convention.