AI norms are in flux---Chinese leadership exports authoritarian AI globally.

Not surprisingly, nascent global AI governance regimes remain ‘unorganized’, ‘immature’, and ‘fragmented’,3 where a range of state and non-state led initiatives and different actors have been competing for influence.4As an ambitious AI player, China has signalled strong interests in shaping, if not making, the emerging global AI order. With the open ambition of becoming a global AI superpower by 2030,5 China is envisioning its leadership role in shaping AI’s future.

This article provides a first in-depth analysis of China’s role in emerging global AI governance. It argues that China’s search for leadership in global AI governance is driven by not only domestic regulatory needs but also the desire to gain international norm and agenda setting power. China’s leadership ambition in global AI governance lies in the wider context of its aspiration to shift from a norm-taker towards a norm-shaper, if not maker. However, despite considerable efforts taken so far, this article suggests that realization of China’s leadership ambition in AI governance is facing enormous challenges especially when nascent global AI governance regimes under the influence of geopolitics and liberal democratic values have in effect allowed China limited room to demonstrate its leadership credentials.

This article proceeds in four subsequent parts. The first part discusses the wider context of China’s shifting views towards global governance and how global AI governance sits in this context. The second part discusses the practical needs for AI governance in the Chinese context. The third part describes China’s current practices and goals in AI governance. The fourth part analyses the challenges that China is facing to realize its AI leadership, followed by concluding remarks.

The Shifting Chinese Views Towards Global Governance and Implications for AI

China’s strong interest in global AI governance lies in the wider context of its shifting views towards global governance. Since the People’s Republic of China was founded in 1949, the then leader Mao Zedong chose to align with the socialist bloc and later the Third World. At the time, international system was viewed as dominated by the West due to be subverted by communist revolutions.6 Not surprisingly, Mao’s China was very much isolated from international community, and its priority was to focus on how to challenge rather than work with global governance institutions.

Ever since Deng Xiaoping took power in the late 1970s and early 1980s, China’s main economic agenda was to take advantage of the global market to develop its economy. This policy approach required much more cooperative relations with the West. China had thus chosen to integrate with international community and increased its engagement with global governance regimes.7 Nonetheless, the engagement was still limited and selective as China’s foreign policy was guided by Deng Xiaoping’s ‘keep a low profile and never claim leadership’ principle.8 The underlying logic of this principle was based on China’s identity as a developing country whose priority should be focusing on domestic rather than international agendas. This principle was also adopted to avoid leadership competition with the US at the global stage so as to maintain a favourable external environment for China to focus on more pressing domestic development.9

Despite the rapid growth of Chinese economy, Deng’s successors Jiang Zemin and Hu Jintao had largely followed the principle. Under their terms, China had selectively engaged in global governance and fended off substantial global responsibility.10 This inactive Chinese approach towards global governance had often led to criticism over China as a ‘free-rider’11 or ‘irresponsible power’12 at the time. To some extent, this had partly contributed to China’s ‘latecomer’ status in global technology governance, as will be discussed later.

Scroll forward to 2008, the financial crisis became a critical moment for China to reflect upon its role in global governance. China’s relatively successful economic performance during the crisis had suddenly increased the world’s expectation of its global responsibility. For example, the concept of ‘G2’—a vague idea advocating China and the US, the two most influential countries, to work together to solve global problems13—had received considerable attention during the crisis. At the time, China openly rejected the idea of G2 as China was neither ready to take any substantial responsibility nor willing to shift its primary focus from domestic issues to international ones, especially when those international problems were considered as a product of the West.14

Nonetheless, this crisis intensified China’s domestic foreign policy debate over whether ‘keep a low profile and never claim leadership’ principle was dated or not, which is one of if not the most important foreign policy debates in contemporary China. Given the rapidly shifting international political landscapes, Deng Xiaoping’s ‘keeping a low profile and never claim leadership’ foreign policy principle as a status quo strategy was increasingly challenged. Opponents argued that China’s role and international political landscapes had become fundamentally different from that when the principle was introduced by Deng decades ago.15 China’s huge economic size no longer allowed it to keep a low profile, and it would no longer be effective in the new circumstances.16If China continues to claim no leadership, its passive attitude would only harm its national interests.17

It was also argued that China’s inactive participation in global governance would only keep China marginalized from the center of power. China would thus have to accept unfair rules and norms made by the West.18 As such, opponents called for a shift from the previous ‘keeping a low profile’ principle to a new ‘striving for achievement’ principle that advocated China to take on more global responsibility and leadership.19 Many Chinese strategic analysts viewed the financial crisis in particular as an excellent opportunity that China could not afford to miss as it created a power vacuum.20 This ‘striving for achievement’ principle noticeably marked a new departure from the original ‘never claim leadership’ principle.

Proponents, however, argued that China’s main challenge was still domestic, and it would be counterproductive if China became too active at the global stage.21 In addition, despite the financial crisis, China’s power gap with the US was still very wide.22 As such, China should continue ‘keeping a low profile’ principle in order to avoid major confrontation with the US. Otherwise, an unfavourable external environment would not be helpful to China’s domestic development.

Nonetheless, even proponents agreed that China should take more responsibility. Otherwise, a global governance system without major global players such as China may dysfunction again and thus lead to another global crisis.23 Therefore, in order to maintain a favourable international environment, China should be more active in participating in global governance. While the debate over China’s ‘keep a low profile and never claim leadership’ principle has not been settled yet, a consensus that China should increase its participation in global governance has gradually emerged. Chinese scholars thus have been discussing how China should not only increase its participation in the existing global governance framework but also promote new China-led secondary rule systems24—an interesting discussion that sheds light on potential Chinese approaches to engage with the global AI governance.

Such shifting Chinese thinking towards global governance has become evident since Xi Jinping took power in 2012. Instead of ‘keeping a low profile and never claim leadership’, ‘striving for achievement’ principle has clearly guided China’s moves on the global stage. Xi has openly called for China to ‘lead the reform of the global governance system.’25Departing from the previous aim for domestic development, the new principle focuses more on promoting political support and leadership at the global stage.26 In order to put it into practice, China has taken considerable effort in reshaping global order in different areas from global economy, climate change to cyber space.27 While China’s actual conduct in global governance is complicated and multifaceted,28 its overall strategic aspiration is to move from norm-taker to a norm-shaper, if not maker.

The above context provides the background for China’s active search for its role in global AI governance. The current Chinese enthusiasm in shaping AI’s future is indeed guided by the new ‘striving for achievement’ principle, which advocates a proactive approach to contest for leadership at the global arena. Global AI governance is considered as a promising area to pilot China’s norm-maker role. In areas such as global economic governance, international norms and institutions such as IMF and World Banks have been established for decades. Thus, it has been very challenging for China to re-shape them despite China’s growing economic influence—not even to mention considerable criticisms over China as a ‘revisionist power’.29 Global AI governance, however, is considered as a new and emerging area where norms and institutions are yet to be created.30 In other words, this leaves more strategic room for China to fill in and realize its leadership credentials. In this regard, China has been planning for early moves to gain an advantageous position in global AI norm and agenda setting.

Not surprisingly, many Chinese scholars are paying close attention to the impact of AI’s rapid development on international order and national security.31 Echoing the aforementioned ‘striving for achievement’ principle to take more global responsibility, there is a general consensus among Chinese scholars that China, as a responsible global power, should be more active when it comes to global AI governance.32 The discussion has thus focused on how to develop and contribute ‘Chinese solutions’ and ‘wisdom’ to tackle governance challenges brought by AI at the global stage.33 Some Chinese research institutions such as the Institute for AI International Governance (I-AIIG) at Tsinghua University have thus emerged as the times require to provide ‘intellectual support’ to China’s participation in global AI governance.34 Emergent joint efforts are also being made by both academia and corporate sectors to facilitate inter-disciplinary and inter-sector collaboration; for instance, Alibaba and Shanghai Jiaotong University have set up a program together specifically focused on AI governance and international rule of law.35

Why Does AI Need to Be Governed?

In the broader context of China’s shifting views towards global governance, the rapid development of AI has created a series of governance challenges that the Chinese government has to tackle. A notably significant issue concerns the impact of AI on the human labour market.36 The rise of AI is likely to restructure the workforce and thus result in massive unemployment and widening socioeconomic inequality.37 The McKinsey Global Institute estimates that the widespread application of AI and automation will force 400–800 million people to change their jobs by 2030.38 Compared with China’s AI near peers such as the US and Europe, China’s relatively low labour and production cost have given its private and public sectors less financial incentives for automation. However, up to 100 million Chinese are still estimated to face employment change by 2030, making China the most affected country during this transformation.39 If not handled well, the potential massive unemployment may threaten social stability at large. Given that social stability is a critical pillar of the Chinese Communist Party (CCP)’s legitimacy,40 this will present a critical challenge to its rule.

In addition, this labour market transformation will not only lead to considerable job loss but also deepen the already widening socioeconomic inequality. The automatization process is likely to benefit the capital that owns AI technology and the advanced skilled workers while reducing the value of low and medium-skilled human labour. According to Organisation for Economic Cooperation and Development (OECD)’s investigation on AI-related online job posting, the skill sets demanded for AI-related workers include technical skills such as python, machine learning, data mining and robotics, and also favourably socio-emotional skills such as communication, teamwork and problem-solving skills.41 In this case, those with low skills are likely to lose in the shifting job market. In other words, the gap between the rich, who own capital and advanced skills, and the poor, who primarily rely on labour value, is likely to get wider in the age of AI. This will present short and mid-term challenges to the CCP who justifies its rule by being a vehicle to deliver a communist society where everyone has equal access to material goods.42 Thus, how to govern and mitigate AI-related risks in labour market requires close attention of the CCP.

Moreover, it is likely to result in widening the existing divides and inequalities not only among individual workers but also countries.43 While advanced technology is always pioneered, developed and controlled by powerful countries and influential companies, there is a high degree of job augmentation and added capacity among developed countries, and meanwhile, the less powerful countries and groups may be placed at the disadvantaged end. Therefore, all countries should pay great attention to the emergent problem, regulating the ‘winner takes it all’ dynamic of AI.44 In this regard, socioeconomic inequality led by the rise of AI is not only domestic but also international, requiring a global solution. Indeed, a PwC prediction shows that China and the US two countries alone will own 70% of the total $16 trillion GDP growth contributed by AI industry.45 As Kai-Fu Lee points out,

“The AI world order will combine winner-take-all economics with an unprecedented concentration of wealth in the hands of a few companies in China and the United States.”46

This future AI order dominated by the US and China scenario is taken for granted by many Chinese scholars. Many thus argue that, given China’s significant role in the future AI order, it needs to work with international society to provide public goods in the field of AI governance.47 Some even argue that China’s AI development has a ‘special mission’ in world politics,48 and China should contribute to tackling the issue of widening global North-South digital divide.49 Given that AI will further widen the technological gap between developed and developing countries, developing countries may lose the last chance to develop itself through economic globalization.50 So the argument goes, China, as one of the very few if not the only developing countries who are capable of becoming a major AI power, is the ‘hope’ of developing countries, and thus China’s AI innovation is to seek developmental rights for not only itself but also the vast number of developing countries.51 It is argued that China should be the forerunner in AI, providing open, shared AI technologies and application platforms to serve the global community.52

Another major concern for governance is AI’s impact on data security and privacy. Arguably, China’s rapid progress in AI innovation is built on its data advantage. As revealed in the 48th Statistical Report on China’s Internet Development, there are over 1 billion Internet users with a penetration rate of 71.6% in 2021.53 Such vibrant market is quite ideal for data collection and algorithm training. However, some predatory data collection has seriously violated privacy and individual rights, which requires immediate regulatory actions from the state. The rapid development and application of facial recognition in particular has been concerning not simply because of the empowerment of state surveillance. With the widespread use of AI technology, there are growing criticisms among the public. Nearly 80% of the Chinese respondents of a survey expressed their concern over data leakage, and more than 70% regarded it necessary to set up specific regulation body to monitor the use of facial recognition technology.54

In addition, AI is a double-edged sword for data security and thus privacy. While many companies rely on AI-enabled solutions to detect and mitigate cyber threats, it also invites unintended security risks. For example, SenseNets, a Chinese company that provides AI-based security protection, was reported to leak confidential personal information—possibly including facial recognition data—involving up to 2.56 million users.55 In response to those issues in data security and privacy, the Chinese government has been exploring ways for regulation and supervision, as will be discussed later.

Moreover, weaponization of AI has drawn considerable discussions nowadays. Some argue that AI will launch a third revolution in warfare, following the first one by gun powder and the second one by nuclear weapons.56 Future AI-enabled weapons are likely to be ‘at least equally impactful’ as nuclear weapons, aircraft and computers.57 Therefore, many countries have made heavy investment in exploring AI’s value in warfare. With the ambition to build a ‘world class’ army, technology plays a critical role in the Chinese army’s ‘modernization’ plan, and AI is a critical emerging technology that the Chinese army is determined to master.

Indeed, to strengthen China’s national defense capacity is a key objective of the Chinese State Council’s AI plan.58 As the plan points out, “through AI, (China should) elevate national defense strength and assure and protect national security.59 This plan also calls to establish an institutionalized communication mechanism between China’s defense and civil sectors such as universities, enterprises and research institutes.60

Needless to say, weaponization of AI has brought complicated security implications and thus led to concerns about its potential impact. Given AI’s military potential, the intensifying great power competitions may escalate the rise of autonomous weapons and robotics. As a report on global military AI market shows, it will grow remarkably and is forecast to reach 26.36 billion by 2027.61 Additionally, the research and development of autonomous weapons and robotics technology is largely concentrated in a few powerful countries.62 Thus, to some, a global AI race is ‘inevitable’.63 China has been closely following how the US has been strategically deploying AI in military use. Likewise, the progress of AI innovation in China and Russia has been frequently highlighted in the US to justify greater urgency and needs to boost similar programs in the US.64 For example, when addressing the Annual Threat Assessment, Director of National Intelligence Avril Haines referred China to an ‘unparalleled priority’ and a ‘formidable challenge’ to the US.65 Such perception has contributed to the rise of securitization trend among great powers where AI has been increasingly labeled as a security matter.66

Among all these issues, the fact that intelligent weapons can operate without human intervention has raised a series of new ethical questions: for instance, to what extent they can be controlled by human, who should be responsible for the killing or unexpected damage due to flawed algorithm, and how to control the manipulation of autonomous weapons from its design to its use. In short, this is about ‘when and if an autonomous system kills, who is responsible for the killing, irrespective of whether it is justified or not?’67 While the character of warfare is transformed from ‘informatized’ to ‘intelligentized’,68 issues in relation to weaponization of AI technology needs to be attentively addressed. All the existing solutions to regulate weaponization of AI, including the human-in-the-loop approach, and ban or regulation,69 require a global dialogue. Many thus argue that global treaties on AI are not only important but also timely.70 According to Fu Ying, a veteran Chinese diplomat, China is highly interested in defining norms to mitigate the risk regarding weaponization of AI.71

Above all, the rapid development of AI has posed a wide range of challenges for governance at both domestic and international stages. While national governments and organizations have made a multitude of strategies, initiatives and plans, a global AI governance framework is required to coordinate a collective response in order to jointly tackle the emerging issues of AI technology and develop safe, trustworthy and controllable AI-enabled systems.

China’s Goals and Practices

Given China’s ambition to be an AI superpower and following its ‘striving for achievement’ foreign policy principle, China is keen to take up leadership responsibility in global AI governance. In order to do so, China has made a series of plans and actions to pioneer AI governance. China’s State Council’s AI plan in 2017 spelt out a timeline and goals for AI governance. According to the plan,

“By 2020, China will be … initially establishing AI ethical norms, policies, and regulations in some areas. By 2025, China will have seen the initial establishment of AI laws and regulations, ethical norms and policy systems, and the formation of AI security assessment and control capabilities. By 2030, China will have constructed more comprehensive AI laws and regulations, and an ethical norms and policy system.”72

Following this ambitious plan in AI governance, China has been actively researching and introducing AI standards and ethics. Progress regarding the protection of privacy and data as well as AI ethics includes but is not limited to the following.

China’s Civil Code was issued in 2020 with specific reference to the protection of privacy and personal information.73

A guideline for self-assessment was drafted in 2020 to propose specific questions regarding the collection and use of personal information by mobile apps.74

New Data Security Law was passed in 2021 to enhance data security and individual and organizational rights.75

The White Paper on the Mobile Apps Personal Information Protection and Governance was published in November 2021, providing the necessary steps for improving governance mechanisms of mobile apps.76

The Personal Information Protection Law was issued in 2021 to set strict rules over personal information collection and use.77

The Guide to AI Ethics was proposed in 2020 with reference to the research, design, application and usage of AI.78

The Ethical Norms for New Generation AI was released in 2021 to incorporate ethics into AI development and to provide ethical guidance.79

The White Paper on Trustworthy AI was published in 2021, highlighting the importance of fairness, accountability and transparency of AI development and giving advice to the government, companies and industries on how to make AI trustworthy.80

All of these progress have signalled the Chinese government’s determination in protecting AI-related privacy, security and ethics. They are indeed made in the background that China’s law in privacy protection and AI ethics tends to significantly lag behind its AI near-peers such as Europe, and thus the Chinese government has been adopting many regulatory practices of its AI near-peers. In other words, China has largely—or is on the way to—endorsed the similar AI-related regulations and laws with that in Western democratic societies.

A primary difference, however, is that Chinese regulatory actions mainly focus on non-state use of data, while those in the West regulates not only non-state but also state actors, which reflects a crucial difference in cultural and political values. For instance, while American AI policies are to a large extent influenced by multiple stakeholders such as political factions, enterprises and research groups, a ‘strong government’ plays an important role in China’s AI policy making.81 In China, based on the assumption that collective, national interests—often defined by the state—should prevail individual citizen rights if there is a clash, the state is thus granted rights to, for example, access ‘data affecting national security, national economy, people’s livelihood, and public interest’ according to its Cyber Security Law.82 Based on its liberal democratic values, Europe’s General Data Protection Regulation (GDPR), however, does not prioritize state interests over individual citizen rights. This different understanding of the relations between individual civil rights and state power provides a basis for a value competition over AI governance that may restrict China’s influence, as will be discussed later.