China's AI law: will we have it soon, or what will we have?
China's been debating its AI law for nearly a year now, but it doesn't look like we'll see a comprehensive AI law like the EU's any time soon. Back in June 2023, China's State Council said they were planning to include AI legislation in their 2023 agenda and even prepped a draft for the National People's Congress to review. However, by September, the 14th National People's Congress Standing Committee's legislative agenda didn't even mention the AI law.
The past year has seen a significant shift in how China's policymakers, legal experts, and tech folks talk about AI regulation. Initially, there was a strong push to regulate, especially after concerns about the security risks of generative AI. But as China's economy started to slow down, and as they began to fall behind in AI tech, particularly in the field of LLM, compared to the US, there's been a lot of second-guessing about whether past regulations might have stifled innovation and hurt China's competitive edge, which in turn could be worsening the economic downturn. Now, the conversation has moved from "What kind of AI law do we need?" to "Do we even need an AI law?"
While it seems unlikely that China will soon enact a sweeping AI law akin to the EU's framework, the discussions are far from over. Given the significant societal impact of AI and China's centralized governance model, some form of legislative action appears inevitable. Moreover, international influences like the EU's AI Act and China's ambition to shape global AI governance continue to pressure domestic legislative developments.
Just last Saturday, the China Artificial Intelligence Industry Alliance (AIIA) held a seminar that drew in folks from government, academia, and even US tech giants like Microsoft and Intel. As expected, the experts from the enterprises raised some common questions, such as pre-trained corpus, copyright, and high-quality datasets. What was interesting was the presence of officials from the National People's Congress Legal Affairs Committee and the Ministry of Foreign Affairs – bodies that you wouldn’t expect to be so involved this early in the process.
It is well known that the Ministry of Science and Technology will lead China's artificial intelligence law, and the participation of the Cyberspace Administration of China as the main responsible department for the interim regulations on Generative AI is also expected. However, the National People's Congress Legal Affairs Committee has always played a relatively silent role in many previous discussions because, according to China's legislative tradition, they usually appear in the second half of legislation (after the responsible government department has formed a draft). The NPC’s Legal Affairs Committee representative walked to the forefront and was listed first in the AIIA Press Release, which is quite intriguing. The Arms Control Department of the Ministry of Foreign Affairs is responsible for representing the Chinese government in multilateral and bilateral negotiations regarding the governance of AI, and naturally should understand the legislative progress of AI as the basis of China's “state practice” and international standpoint, but such early participation is also not very usual.
This tells me that China's not about to drop AI law completely. But will it be as a comprehensive AI law like the EU’s AI Act, as originally thought? That perhaps remains to be seen.
Promoting Development Through Governance, Advancing Goodness in AI—A Successful Symposium on Major Issues in Artificial Intelligence Legislation
On April 19, 2024 afternoon, the "AI Good Governance Forum - Industry Seminar on Major Issues in Artificial Intelligence Legislation" was successfully held in Beijing. The conference was hosted by the Policy and Regulation Working Group of the China Artificial Intelligence Industry Alliance (AIIA) and organized by the Institute of Policy and Economics of China Academy of Information and Communications Technology. Experts from the Legal Affairs Committee of the Standing Committee of the National People's Congress, the Cyberspace Administration of China, the Ministry of Foreign Affairs, the Ministry of Science and Technology, and other bodies, as well as experts from universities and research institutes such as China University of Political Science and Law, Beihang University, East China University of Political Science and Law, Southwest University of Political Science and Law, Zhongnan University of Economics and Law, Northwest University of Political Science and Law, Chinese Academy of Sciences, and the China Institute of Science and Technology Information, attended the conference. Representatives from more than 20 companies such as Baidu, Douyin, JD.com, Ant, Microsoft, Alibaba, and Intel attended the conference. Experts and business representatives engaged in a lively discussion on "establishing an inclusive, prudent, and effective governance of artificial intelligence, and formulating laws that promote economic innovation and foster inclusive sharing."
Wang Zhiqin, Vice President of China Academy of Information and Communications Technology: In his opening speech, he pointed out that artificial intelligence is a strategic technology that leads a new round of technological and industrial revolution, and is also an important engine for new quality productivity. Major economies such as the US and the European Union are constantly accelerating the governance of artificial intelligence technology, and the international community is also actively exploring international governance plans for artificial intelligence. Domestic artificial intelligence legislation should keep up with the times, guide technology to be "people-oriented" and "intelligent for good", and also adapt to the reality of China's industry, fully considering the interests and expectations of all sectors of society.
He Wei, Deputy Chief Engineer of China Academy of Information and Communications Technology , presided over the "Artificial Intelligence Industry Status Quo and Legislative Thinking" theme forum and invited five enterprise representatives to speak.
Baidu's Vice President of Legal Affairs, Wu Mengyi, delivered a speech titled "Reflections on Artificial Intelligence and Its Governance in the Era of Large Models", highlighting three major challenges facing the industry: lack of data quantity and quality, unclear definition of data infringement liability, and insufficient protection of content rights. Wu Mengyi emphasized that the rule of law is the best protection for technology companies. She expressed hope that new legislation would establish a new order of rules, further reduce market transaction costs, and thus enhance the productivity of enterprises.
During the symposium on "Legal Challenges and Governance Prospects of Artificial Intelligence," Tai Jiangli, Vice President of Legal Affairs at Douyin Group, focused on the diverse development of AI applications. She pointed out the weaknesses in the quantity and quality of Chinese language corpora. Tai emphasized that overly restrictive copyright protection and personal information usage regulations exacerbate the difficulties enterprises face in data acquisition, and she expressed hope for increased involvement from relevant parties in legislative discussions.
Subsequently, He Zhongkai, Vice President of the Legal Research Institute at JD Group, spoke on the theme "Adhering to Problem-Oriented Legislation to Promote the Development and Application of Artificial Intelligence." He stated that artificial intelligence, as a new productivity tool, needs guidance for positive development. Legislation should balance the rights and interests of developers, providers, and users while addressing the diverse challenges faced by enterprises in platforms, assistive diagnostics, logistics distribution, and commodity marketing.
Li Zhenhua, President of the Research Institute at Ant Group, delivered a speech titled "Legislative Considerations for Generative AI Applications." He highlighted the necessity of creating a relaxed environment for the industrial applications of large models, encouraging applications in finance, healthcare, government services, and public welfare. Additionally, Li suggested increasing the digitization rate of offline structured data such as papers and books to exempt training data from copyright restrictions, moderately exempting the use of publicly available personal data, accelerating the opening of public data, and possibly facilitating the construction of public training datasets by relevant departments to be opened to some major model manufacturers.
Following the theme forum, the conference transitioned into the first discussion session. Zhang Chunfei, Deputy Director of the Policy and Economy Research Institute at the China Academy of Information and Communications Technology and Co-Chair of the AIIA Policy and Regulation Working Group, summarized the demands of the representatives from various companies and the trends in industry development. He also briefly introduced the preparatory work done by the Policy and Regulation Working Group for this meeting.
Zhao Jingwu, Associate Professor at the School of Law, Beihang University, and Deputy Chair of the AIIA Policy and Regulation Working Group, summarized the legal compliance pain points and difficulties in artificial intelligence. The working group collected over 50 suggestions from companies, covering aspects such as lawful use of data, obligations for data security, use of open-source models, copyrightability of generated content, and coordination of registration systems. The suggestions collected highlight areas where current legal frameworks can be improved, many of which resonate with the "Artificial Intelligence Law (Scholar's Suggested Draft)." Zhao Jingwu then categorized the enterprise suggestions into six sections: the concept and scope of AI, the legality of datasets, intellectual property systems, allocation of liability, utilization of open-source models, and institutional coordination. These included principles for allocating "contribution" in copyrightable content and special supervision of industrial AI. The suggestions reflect the real needs of the industry and provide important reference material for future legislative work.
The "Artificial Intelligence Law (Scholar's Suggested Draft)" was released on March 16, 2024. Cheng Ying, Senior Engineer at the China Academy of Information and Communications Technology and Secretary-General of the AIIA Policy and Regulation Working Group, spoke on behalf of the draft committee. The suggested draft summarizes best practices and recent advancements in AI legal governance, aiming to promote the establishment of a robust AI governance system in China. The draft was intended to provide a platform for dialogue and better integrate wisdom from various sectors, consolidating industry consensus. The research on the AI law draft follows principles of setting foundations, outlining the main framework, and leaving room for integration. It boldly addresses and seeks solutions to the pain points of concern to the industry, exploring system highlights such as liberalizing reasonable data use, establishing special systems for critical AI, fostering an open-source ecosystem, differentiating regulatory approaches for specific scenarios, and establishing compliance exemptions. These are intended to be forward-looking, advanced, and locally relevant. Associate Professors Han Xuzhi from East China University of Political Science and Law and Xu Xiaoben from Zhongnan University of Economics and Law added their comments. Han Xuzhi argued that relaxing the rules for reasonable use of data does not imply an increase in legal risks, as torts committed using AI are mainly concentrated at the output stage. A nuanced liability system around model infringement could provide ample space for technological development. Xu Xiaoben suggested that AI legislation should focus on AI services, clarify the scope of adjustment, address core governance needs, and avoid overly broad regulation of AI technologies.
Following the theme forum, the conference transitioned into its second discussion session. The session was chaired by Professor Zhang Linghan, a professor at the China University of Political Science and Law, a senior consultant at the United Nations, and co-chair of the AIIA Policy and Regulation Working Group. Professor Zhang highlighted the importance of participation by AI enterprises of different scales and stages of development in advancing AI legislation, and invited corporate representatives to discuss topics such as "How can the system adapt to technological changes?" and "How to promote AI innovation and international cooperation," expecting participants to raise urgent, pressing, and prominent issues.
Under the theme of "How the system can adapt to technological changes," chief data legal expert from Tencent Research Institute Wang Rong, director of the AI Governance Center at Alibaba Research Institute Fu Hongyu, AI Technology and Standards Director at Intel Wang Haining, and the head of AI business legal affairs at Baidu Xu Quanquan, spoke on their experiences transitioning from algorithm governance in the platform economy to current AI governance.
Wang Rong believes that continuing traditional regulatory approaches is inadequate for the rapidly changing technology landscape, and that guidance-oriented legislation is more reasonable. The choice of governance pathways is significantly influenced by the international competitive landscape. How to push local industries to compete on the global stage with leading international enterprises is a topic worth considering for all parties. Fu Hongyu discussed the challenges to information security governance posed by the operating principles of large models, noting that directly filtering and intervening in generated content is difficult. A regulatory approach that emphasizes strict rules and strong supervision would impose considerable compliance costs on enterprises. AI legislation should accommodate the uncertainty of technological development, focusing on incremental risks.
In the discussion on "How to promote AI innovation and international cooperation," enterprises that have emerged following breakthroughs in AI technology participated. The Director of AI Ethics and Governance at SenseTime, Hu Zhengkun, MiniMax Vice President and Editor-in-Chief Peng Tao, Vice President of Public Affairs at Shenyin Technology Wang Yujie, the legal head of an AI company Wu Shaoqing, and the legal expert from Baichuan Intelligence Ge Yichen each made statements.
Hu Zhengkun noted the lack of a systematic approach to AI governance in China, which causes communication barriers when participating in international governance. When local AI governance concepts go global, they face a lack of discursive power, affecting the smooth internationalization of products. The next step must consider how to streamline domestic governance systems and expand international influence. Peng Tao, considering the challenges faced by domestic AI companies going international, suggested establishing an international think tank for global AI governance to provide specialized compliance guidance abroad; and supporting outstanding corporate representatives in computing power, corpora, funding, and policy to enhance their international competitiveness. Wang Yujie focused on the technical talent aspect, pointing out the scarcity of international talent in technology and governance for companies going abroad. Legislative and policy attention should be maintained on the introduction of foreign talent and international expert exchanges. Wu Shaoqing affirmed the views on promoting international cooperation and talent mobility and proposed establishing a dual-layer ethical structure for technology: the first layer accommodating universally recognized values and the second developing localized values. Ge Yichen believes that AI innovation and international cooperation cannot be separated from technological competitiveness and discursive power. The former relies on domestic systems to fully promote technological development and cover the entire industrial chain with unified responsibilities; the latter comes from a governance philosophy that is rooted in practice and maintains openness.
This conference provided a platform for corporate representatives and academic experts to report on the current state of the AI industry, summarize the characteristics and patterns of AI industry development at home and abroad; discuss the positioning and goals of AI legislation, and refine the difficulties and questions faced in drafting provisions; and thoroughly discuss the basic concepts and implementation paths of AI governance, providing important references for China's AI legislation and practices, and playing a positive role in promoting the healthy development of China's AI industry and aiding the development of Chinese AI legislation.