Abstract
China's progress in artificial intelligence (AI) is remarkable. It has rapidly become a global leader in AI research, producing a significant volume of publications and patents. Additionally, China has excelled in the commercialization of AI, especially in sectors such as speech and image recognition. The development of a digital and AI-related economy has dramatically increased the size of the gig economy, with an official estimate of approximately 200 million workers in 2021, 1 among which 84 million were in new forms of employment such as food delivery riders, couriers, and ride-hailing drivers. 2 Most recently, the country has set itself apart by introducing some of the world’s earliest and most comprehensive regulations concerning algorithms and platform work, deepfakes, and generative AI, marking it as a front-runner in AI regulation compared to other jurisdictions. How does China regulate AI? And what are the impacts on the Chinese workplace?
Through an examination of major national-level AI-related policies and regulations through early 2024, especially focusing on their implications for the workplace, we contend that the Chinese Party-State adopts an instrumental approach in crafting and implementing AI policies and regulations. In the work and employment arena, this approach seeks to strike a balance between the objectives of efficiency and social stability. Although public input occasionally influences labor-friendly regulations, the predominant emphasis on economic growth over labor protection leaves workers vulnerable to the impact of AI, tech firms, and the overarching authoritarian legal and political framework.
The Chinese Model of Permissive and Restrictive Regulation of AI
Many observers attribute China’s remarkable AI development to the strong supportive policies of the Party-State. Therefore, China’s shift toward comprehensive and, in many aspects, stringent AI regulations, alongside a crackdown on its AI tech giants since the late 2010s, has surprised the world. Does the Party-State intend to stifle China’s innovation? Will AI be completely controlled and monitored by the Party-State? To fully understand AI governance in China, it is crucial to consider the AI policies and regulations within the context of China’s evolving political economy. Since the economic reform, the Party-State has consistently sought to balance two interrelated yet potentially contradictory objectives that underpin regime legitimacy: economic development and social stability (Lee 2007; Dickson 2016). On one hand, economic liberalization may exacerbate inequality and undermine social stability. On the other, robust government intervention and labor protection could stifle investment and economic growth.
Our analysis of China’s major national-level AI policies and regulations, alongside existing studies, reports, and our interviews with Chinese AI and platform companies, trade unions, and workers, suggests that the Party-State embraces authoritarian legality (Lee 2016; Gallagher 2017; Whiting 2023), using AI policies and regulations instrumentally to achieve its objectives of techno-development and the economic, social, and political security of the Party-State. To balance these potentially conflicting objectives, China's AI regulations exhibit several notable characteristics. First, many regulations are issued as agency-level directives, allowing for both arbitrariness and adaptability. Second, these regulations often involve multiple potential enforcers from diverse government agencies and at various levels, each of whom may interpret and enforce the rules differently. This arrangement also affords the central government the flexibility to intervene as needed. Third, many regulations are intentionally ambiguous, facilitating lax, flexible, or selective enforcement. Finally, significant regulations often undergo an exposure draft phase to gather feedback from stakeholders, ensuring that the final regulations are more balanced and reflective of broader interests. This permissive and restrictive approach to AI governance is reflected in the Chinese term “inclusive and prudent regulation,” first introduced by the late Premier Li Keqiang in his Report on the Work of the Government on March 5, 2017, 3 and later adopted as a regulatory principle for the broadly defined new economy, including AI. In particular, Article 3 of the Interim Measures for the Administration of Generative Artificial Intelligence Services specifies the “inclusive and prudent” regulatory principle of AI: “The state is to adhere to the principle of placing equal emphasis on development and security, merging the promotion of innovation with governance in accordance with law; employing effective measures to encourage innovation and development in generative AI, and carrying out tolerant and cautious graded management by category of generative AI services.” 4
China’s supportive policies and tolerant regulations toward AI are both comprehensive and significant, boosting the development of its “AI National Champions”—Baidu, Alibaba, Tencent, and iFlytek—as well as supporting the global expansion of Chinese platform companies such as TikTok, Temu, and Shein. By the late 2010s, however, as major AI and platform companies began to dominate large swaths of data, financial, and social activities and gained significant influence in digital infrastructure, their growing power posed threats to the Party-State’s economic and political security. In response, China shifted its regulatory approach from permissive to restrictive, focusing on curbing the financial services of platforms, strengthening antitrust measures to prevent excessive capital expansion in the AI and platform economy, and imposing greater political and ideological control and monitoring over data and AI usage. Additionally, the Party-State has begun to bolster labor protections in the platform economy to address concerns about social stability. This study details the major workplace AI policies and regulations that align with the Party-State’s goals of efficiency and social stability (see Table 1 for a summary).
China’s Key Regulations, Policies, and Directives on Workplace AI, Algorithm, and Platform Work
Notes: 1. Interim Measures for the Administration of Generative Artificial Intelligence Services; 2. Measures for the Administration of Generative Artificial Intelligence Services (Exposure Draft); 3. Measures for Scientific and Technological Ethics Review (for Trial Implementation) (Exposure Draft); 4. Provisions on the Administration of Deep Synthesis of Internet-Based Services; 5. Opinions on Strengthening the Building of a Highly Skilled Workforce for the New Era; 6. Guidance Training Program for Graduate Students in Artificial Intelligence (for Trial Implementation); 7. Opinions on Strengthening the Governance over Ethics in Science and Technology; 8. Action Plan for Enhancing Nationwide Digital Literacy and Skills; 9. Opinions on Promoting High-Quality Development of Modern Vocational Education; 10. Ethical Norms for New Generation Artificial Intelligence; 11. Several Policies for Promoting the High-Quality Development of the Integrated Circuit Industry and the Software Industry in the New Era; 12. Several Opinions on “Double First-Class” Construction of Colleges and Universities to Promote the Integration of Disciplines and Speed up the Education of Postgraduate Students in the Field of Artificial Intelligence; 13. Guidelines on Promoting the Development of Artificial Intelligence in Forestry and Grassland; 14. Governance Principles for the New Generation Artificial Intelligence – Developing Responsible Artificial Intelligence; 15. Vocational Skills Improvement Action Plan (2019–2021); 16. Educational Informatization 2.0 Action Plan; 17. Artificial Intelligence Innovation Action Plan for Institutions of Higher Education; 18. General Senior High School Curriculum Program and Curriculum Standards for Chinese Language and Other Subjects (2017 Edition); 19. Three-Year Action Plan for Promoting the Development of New Generation of Artificial Intelligence Industry (2018–2020); 20. Development Plan on the New Generation of Artificial Intelligence; 21. Outline of the Innovation-Driven Development Strategy; 22. “Internet+” and Artificial Intelligence Three-Year Action and Implementation Plan; 23. Robotics Industrial Development Plan (2016–2020); 24. Implementation of the Intelligent Manufacturing Development Plan (2016–2020); 25. Several Policies on Further Encouraging the Development of the Software and Integrated Circuit Industry; 26. Several Policies on Encouraging the Development of Software Industry and Integrated Circuit Industry; 27. Notice on Strengthening the One-Stop Mediation for Labor Disputes in New Forms of Employment; 28. Guidelines for Guaranteeing the Rest and Labor Remuneration Rights and Interests of Workers in New Forms of Employment; 29. Guidelines for the Publicity of Labor Rules for Workers in New Forms of Employment; 30. Service Guidelines for Protecting the Rights and Interests of Workers in New Forms of Employment; 31. Guidelines for the Conclusion of Employment Contracts or Written Agreements with Workers Employed in New Forms (for Trial Implementation); 32. Opinions on Providing Judicial Services and Guarantees for Stabilizing Employment; 33. Provisions on the Administration of Algorithm-Generated Recommendations for Internet Information Services; 34. Opinions on Promoting the Standardized and Healthy Development of the Platform Economy; 35. Opinions on Strengthening the Protection of the Rights and Interests of Workers in New Forms of Transportation Industry; 36. Guidelines for Implementing the Primary Responsibility of Internet Platforms (Exposure Draft); 37. Opinions on Strengthening the Protection of the Rights and Interests of Truck Drivers; 38. Guidance on Strengthening the Comprehensive Governance of Network Information Service Algorithms; 39. Guidance on Implementing the Responsibilities of Online Catering Platforms to Effectively Protect the Rights and Interests of Takeout Delivery Personnel; 40. Guidance on Protecting the Labor Rights and Interests of Workers in New Forms of Employment; 41. Guidance for Anti-Monopoly in the Field of Platform Economy; 42. Guidance on Promoting the Well-Regulated and Sound Development of the Platform Economy; 43. Interim Measures for the Administration of Online Taxi Booking Business Operations and Services; 44. Guidance on Deepening Reform and Promoting the Sound Development of the Taxi Industry; 45. Outline of the National Informatization Development Strategy; 46. Digital Skills Upgrading Program for Female Workers; 47. ACFTU’s Actions to Widely Apply Artificial Intelligence; 48. Plan to Accelerate the Digitization of Trade Unions; 49. Three-Year Action Plan to Promote Unionization of Workers in New Forms of Employment; 50. Notice on Promoting the Construction of a Coordination and Consultation Mechanism for the Rights and Interests of Workers in New Forms of Employment; 51. Guidance on Further Advancing “Grassroots Unionization” Work; 52. Amendment to the Trade Union Law; 53. Several Opinions on Promoting the Unionization of Workers in New Forms of Employment; 54. Program to Promote the Unionization of Truck Drivers and Other Groups.
Supporting and Regulating AI for Efficiency
Believing that the information and communication technology (ICT) sector is pivotal for China’s economic growth and technological advancement, the Party-State began fostering a supportive policy environment for internet companies as early as the late 1990s. This effort intensified after the 2008 global financial crisis, when the internet was designated as a national pillar industry essential for the country’s industrial upgrading and “indigenous innovation” (Lei 2023; McKnight, Kenney, and Breznitz 2023). Around the mid-2010s, the Party-State began to support AI development, as evidenced by national policies such as Made in China 2025, the Robotics Industrial Development Plan (2016–2020), and the “Internet+” and Artificial Intelligence Three-Year Action and Implementation Plan. In 2017, China published the Development Plan on the New Generation of Artificial Intelligence, which positions AI as the primary driver of China’s industrial upgrading and development, outlines its strategy to center AI in its socioeconomic development efforts, and articulates its ambition to become a global leader in AI with an industry worth approximately $1.38 trillion by 2030. Since then, various support policies and accommodative regulations have been implemented to promote AI development, focusing on key areas including talent, data, capital and market access, and hardware. Next, we examine two aspects relevant to the workplace: talent/skills and employment flexibility.
Talent/Skills
To cultivate and recruit talented AI researchers and workers, various policies adopt a dual “training” and “gathering” approach. On the training front, China takes a long-term view to nurture AI talent by integrating AI education into the curricula of various levels of general and vocational education. A strong emphasis has also been placed on strengthening international exchanges and cooperation, encouraging joint support from national funds, financial sectors, and societal capital for AI discipline development and talent cultivation. Experimental zones for digital skills training are being established, fostering collaboration between academia, industry, and research sectors to cultivate AI talent. Moreover, infrastructure development for AI education such as gigabit optical networks, 5G networks, and Internet Protocol version 6 (IPv6) is being prioritized. Additionally, China is focusing on orderly and legally opening up public data resources and improving digital literacy and skills for citizens of all ages and demographics. On the gathering front, numerous national- and local-level talent programs have been developed to attract AI talent to work in China. Additionally, China’s AI companies have established their own overseas institutes to recruit international talent, further bolstering the country’s AI development capabilities.
Employment Flexibility
Through regulatory tolerance and exemptions, China provides its platform companies with much-desired employment flexibility. In particular, the Interim Measures for the Administration of Online Taxi Booking Business Operations and Services (2016), promulgated by the Ministry of Transport and six other government agencies, exempt ride-hailing platforms from establishing an employment relationship with their drivers. Additionally, the Guidance on Promoting the Well-Regulated and Sound Development of the Platform Economy, issued by the General Office of the State Council in 2019, encourages the expansion of new forms of employment. These developments suggest that platform companies may enjoy at least partial immunity from legal liabilities related to employment issues. Consequently, Chinese platforms have shifted their employment model from direct employment to two more flexible arrangements that bypass a formal employment relationship: subcontracting (i.e., labor dispatching) and crowdsourcing (Lei 2021, 2023). Despite calls from the National People's Congress and the Chinese People's Political Consultative Conference for the Ministry of Human Resources and Social Security (MOHRSS) to recognize the employment relationship between platforms and their workers, MOHRSS has rejected such classifications, arguing that they would harm the platform economy and employment opportunities. For similar reasons of efficiency, courts have rarely recognized the employment relationship between platforms and their workers, even after the extensive regulatory crackdown on AI and platform companies in 2020 (Lin 2022; Zhang 2024).
Regulating AI for Social Stability
Maintaining social stability has been a key objective for the Party-State since the economic reform began in 1978 (Lee 2007; Liu and Kuruvilla 2017). The platform economy, characterized by a centralized and monopolized market structure and ambiguous employment relationships (Tirole 2017), presents different and potentially more severe challenges to social stability than traditional sectors. First, platform workers, who often operate outside formal employment relationships, are more inclined to engage in collective actions to address their labor disputes. This is largely because they face significant difficulties in accessing legal channels for dispute resolution. Second, unlike traditional cellular labor activism (Lee 2007), platform workers tend to protest against the same companies or different companies over similar grievances, which facilitates the building of solidarity across workplaces or regions. Third, since many platform workers are involved in transportation services, their protests frequently disrupt traffic and public order, 5 posing a greater risk of causing social instability (Lei 2023).
The risk of labor unrest in the platform economy prompted the Party-State to adopt a cautious approach in formulating regulations, with a focus on three areas: AI and algorithms, labor standards, and employee voice. In regulating AI and algorithms, China mandates algorithm registration and transparency regarding work rules and labor standards, establishes AI ethical review and regulatory systems, supports third-party assessments of algorithms, and requires platforms to establish AI accountability mechanisms and prevent prejudice and discrimination.
Regarding labor standards, the regulations require platform companies to avoid discrimination in hiring, adopt a “middle-ground algorithm” for work assessment, pay non-standard workers local minimum wages, inform ride-hailing drivers about commission rates and their adjustments, supervise wage payments by partners, provide extra compensation for working on statutory holidays, ensure workers’ rest time, count both online and offline time as working hours, prevent overwork, and establish occupational injury insurance systems for platform workers. Significant labor rule adjustments must be reported to local human resources and social security departments in advance. Platform companies and their partners must clearly define employment responsibilities and sign written agreements with workers in non-standard employment relationships.
The regulations also seek to promote employee voice to reduce labor disputes and conflicts by mandating platform companies to solicit opinions from workers before formulating or revising labor rules, disclose algorithm rules to them, recognize the right of platform workers to join trade unions and negotiate working conditions, affirm the right of trade unions in supervising labor standards compliance, encourage the establishment of internal labor dispute mediation committees, and develop mediation and litigation mechanisms for resolving labor disputes. However, platform workers are banned from using illegal or extreme means to defend their rights.
While these regulations appear comprehensive and stringent, enforcement by the Party-State has been lax. This outcome is partly because, like other labor regulations in China, these new rules are often too vague or unfeasible to enforce effectively. For example, Article 20 of the Provisions on the Administration of Algorithm-Generated Recommendations for Internet Information Services mandates that platforms ensure algorithms provide workers with adequate compensation and rest, yet it remains unclear what legally constitutes “adequate.” More fundamentally, the intention of the Party-State may not be so much about labor protection as about containing labor unrest (Lei 2023). The right of unionization and collective bargaining for platform workers, while theoretically beneficial, is not particularly effective because of the nature of the All-China Federation of Trade Unions (ACFTU)—the sole official labor union that monopolizes worker representation in China.
The Role and Responses of Social Actors
China's AI policies and regulations, although predominantly influenced by the Party-State, also incorporate input from a variety of social actors, including AI and platform companies, trade unions, workers, and the general public. Initially operating in a legal gray area, major AI and platform companies have sought to influence policies and regulations from their inception through direct and indirect lobbying of central and local governments. Their tactics include hiring director-level government officials with lucrative compensation to manage government relations; sponsoring academic research that underscores the platforms’ contributions to economic development, employment growth, and poverty relief; and investing in employment opportunities for various vulnerable groups such as the unemployed, disabled, and veterans. These efforts have contributed to the leniency in platform labor regulations (Zhang 2024).
The ACFTU has been commanded by the Party-State to manage increasing pluralistic interests and counter the emergence of independent labor movements. Despite some innovative and successful reforms within the ACFTU, particularly at the regional and grassroots levels, union organizing and collective bargaining in China have largely remained symbolic due to the tight grip of the Party-State, rendering the ACFTU ineffective at truly representing workers' interests (Liu 2009, 2020; Friedman 2014).
The inability of the ACFTU and the legal system to resolve labor insurgency has increasingly led the Party-State to adopt a strategy of repression to manage labor unrest (Chen and Gallagher 2018; Li and Liu 2018). The rise of worker-led collective bargaining, supported by movement-oriented labor nongovernmental organizations (NGOs), has particularly challenged the Party-State, leading to a severe crackdown on labor activists and labor-friendly NGOs since 2015. This repression has limited the space for civil society and independent worker mobilization, coinciding with the development of AI and platforms in China, providing a crucial context for understanding China's AI governance and its workplace implications.
The ACFTU plays a significant role in platform labor regulation, especially at the national level. By early 2024, the ACFTU has issued five AI-related policies/regulations in conjunction with other Party-State agencies, and eight policies/regulations independently (see Table 1). These initiatives focus on labor standards, employee voice, and the rights of unionization and collective bargaining in the platform economy, in addition to the digitalization of trade unions' work to enhance their efficiency.
Adhering to the principle of “where there are workers, there should be trade unions” (Liu, Li, and Kim 2011), the ACFTU has begun to organize platform workers—who often operate outside formal employment relationships—into unions. However, the traditional ACFTU organizing model, which relies on employers’ permission for unionization, is ineffective in the context of online platforms. Consequently, alternative models such as local industry union associations and regional union associations are being utilized (Liu 2010). In some instances, these groups engage in collective consultation on working conditions. Although these union organizing and bargaining models can rapidly increase union membership among platform workers, the resulting unions and negotiations tend to be largely symbolic, bringing minimal benefits to workers. For example, Lalamove in Shenzhen established a company-level union in 2019 that facilitated communication between drivers and the platform and helped increase task rewards and subsidies. It did not address the drivers' core demand for higher commission rates, however. 6 Our interviews conducted with platform company managers, workers, and union staff revealed that none could identify any substantial gains for workers from these unionization efforts. Additionally, when dissatisfied platform workers have attempted to organize their own groups, they have faced detention by the police, indicating significant challenges to genuine labor organization and advocacy in China. 7
Chinese platform workers often face precarious employment conditions characterized by extremely long working hours, low commission rates, unsafe working conditions, and limited or no access to social insurance. Furthermore, many are technically employed and managed by third parties but do not receive formal labor contracts or social insurance benefits (Lei 2021; Liu and Friedman 2021). This hyper-exploitation has led to a rise in both individual and collective protests among platform workers. Despite the Party-State's strict ban on independent worker organizing and repression of radical collective actions, the number of collective protests by platform workers remains notable. For instance, the China Labor Bulletin recorded 90 collective protests by couriers, food-delivery riders, ride-hailing drivers, and truck drivers in 2023 alone. 8 Individual protests by platform workers are also numerous, with some resorting to radical means. Consider several examples: in 2021, an Ele.me food-delivery rider set himself on fire over unpaid wages; 9 in 2022, another Ele.me rider stabbed himself after being threatened with a 1,000 renminbi (RMB) fine if he resigned; 10 and in 2023, a Lalamove driver fatally stabbed two platform staff members after he was unable to retrieve his deposit of RMB 1,000 upon resignation. 11 These protests of platform workers present unique challenges to social stability, prompting the Party-State to strengthen labor regulations in the platform economy.
The role of journalists, academic researchers, and the general public has been crucial in advocating for labor-friendly regulations in the platform economy. A striking example occurred in September 2020 when the Chinese magazine Renwu published a report that vividly illustrated the challenges faced by food-delivery couriers. These workers' schedules and incomes were dictated by algorithms that failed to consider the realities of navigating Chinese cities. The publication of this report sparked widespread public outcry, which pressured platform companies to re-assess their practices and accelerated the implementation of new labor regulations.
The impact of these regulations on the workers themselves, however, appears to be extremely limited, if any. In interviews we conducted with food-delivery couriers and ride-hailing drivers, the vast majority reported no positive changes in their working conditions or in the resolution of labor disputes. Instead, many expressed frustration that new platform rules had actually reduced their earnings. For instance, some workers criticized regulations that mandate rest periods after a certain number of hours of work, arguing that these do not alleviate fatigue but instead decrease their income. Additionally, despite Meituan and Ele.me making their algorithms public, none of the couriers we interviewed were aware of the details. For these workers, changes that do not directly lead to increased commission rates are perceived as insignificant.
From the management side, platform managers reported that the new labor regulations have not significantly burdened their operations. Some regulations, such as mandatory rest periods after certain hours of work and the establishment of an occupational injury insurance system, even help platforms reduce work-related injuries and the associated compensation costs. This discrepancy between the experiences of workers and the perceptions of management highlights a critical gap in the effectiveness of the labor regulations. It suggests that while the regulations may be designed to improve working conditions and fairness, their actual implementation and the tangible benefits they provide to workers might not align with these objectives.
Conclusion
The rapid development of AI and platform companies in China has led to explosive growth in gig employment, posing challenges to social stability and regime legitimacy. To address issues facing platform workers—such as hyper-exploitation by algorithms, lack of social insurances, and limited access to labor dispute resolution—China has implemented significant regulations on workplace AI and platforms. A key characteristic of China's regulatory strategy is its strategic ambiguity, which grants regulators considerable flexibility. This flexibility, however, has led to issues such as lax and selective enforcement. Consequently, evidence of substantial improvements in labor rights within the platform economy is yet to be seen. While China’s approach to AI workplace regulation oscillates between permissive and restrictive, aiming to balance efficiency with social stability, it often exposes workers to the harsh realities of techno-capitalism under an authoritarian regime. This situation underscores a complex interplay between technological advancement and labor rights, reflecting the inherent tensions in China's AI governance through which technological progress is aggressively pursued at potential social costs. The critical question remains: Can China reconcile these tensions in the long term and balance rapid technological development with its broader societal impacts?
