China's AI Policies: Overview and US Comparisons

Introduction

As of 2026, China and the US have become the two largest poles of AI development. Both countries share ambitions of leading broad AI deployment across industries and shaping the technology’s global trajectory, backed by significant investment. However, the paths they are taking differ significantly — particularly in the relationship between private companies that act as national champions and state institutions.

The US approach is shaped by public narratives that treat generative AI as a technology of unprecedented novelty, often understood best by the companies developing it. That framing tends to leave government in a more reactive role, channeling an industrial strategy directed primarily by large private actors and relying heavily on voluntary commitments, with a near-exclusive focus on results over process. In contrast, China has inscribed its AI policy in a direct continuation of its ongoing digital policy. The state takes a more active role in directing infrastructure and research investments, encouraging cooperation between private actors, adapting existing legal frameworks to new technical paradigms ex ante, and leveraging state-owned enterprises and other public instruments to push adoption and fill gaps where private capital is hesitant — both domestically and internationally.

In 2025 and 2026, China issued several documents expanding and operationalizing this approach—such as the 15th Five-Year Plan (2026–2030), which sets development goals for the next five-year period. These documents organize strategy around self-reliant development of new AI systems through infrastructure, computing power, and data advantages, alongside facilitating uptake of AI applications across economic domains within existing regulatory frameworks. Data governance and cross-border cooperation stand out among the levers that structure these strategies.

The present report provides an overview of recent regulatory and organizational developments (through April 2026) supporting China’s AI strategy along these lines. Part I reviews the main policy documents and concepts at play. Part II focuses on fostering the development and provisioning of resources for large-scale generative AI—including data, energy infrastructure, domestically developed compute, and open-source strategy and diplomacy supporting international data flows. Part III addresses how AI developers and deployers are governed through competition law and pre-deployment algorithm and generative-AI filing requirements that carry privacy, data-security, and content rules into model development and launch.

I. Governance Frameworks for China’s AI Strategy

China’s AI strategy framework includes aspects of industrial policy, data compliance, and competition law, as well as processes for leveraging fiscal funds and state-owned enterprises. Expand any item below for a brief introduction to the relevant concept or document.

II. Building up AI: Energy, Compute, and Data

A. Developing Energy and Compute Infrastructure

China’s AI+ strategy Guidance and the 15th Five-Year Plan show a joint focus on computing power infrastructure, algorithmic innovation, and data resources to support AI development. In particular, the 15th Five-year Plan sets up goals to make computing power more accessible and affordable by construction of ultra-large-scale “intelligent computing” clusters, coordinated deployment of green electricity as well as computing capacity, and strengthening nationwide integrated monitoring and scheduling of computing resources. (1)

“Computing power” is defined in the Action Plan for the High-Quality Development of Computing Infrastructure (Gong Xin Bu Lian Tong Xin (2023) No.180) (2) as “a new form of productivity that integrates information processing capabilities, network transmission capacity, and data storage capabilities; it primarily delivers services to society through computing infrastructure.” To meet the energy needs of this new physical infrastructure, China has significantly ramped up its power capacity production across all energy technologies since 2021; its additions in the last five years are estimated to be equivalent to the total US capacity and the government plans to add six times as much over the next five years, (3) with the greatest increase coming in the form of wind and solar energy. (4)

In order to accelerate this development despite the low attractiveness of infrastructure projects to private developers, with their high upfront costs and slow returns, China has injected fiscal funds and developed specific financial instruments to promote AI infrastructure, following recommendations outlined in the AI+ Strategy Guidance to encourage “long-term, patient, and strategic capital”. (5) In November 2025, China allocated a new ¥500 billion ($70.3 billion) policy-based financial instrument to propel mainly tech-driven projects and urban renewal programs. (6) In January 2026, four central government ministries issued a Notice on Optimizing the Implementation of the Fiscal Interest Subsidy Policy for Equipment Renewal Loans (Cai Jin [2026] No. 2) stating that where business entities within certain sectors, including AI, receive loans from banks for equipment renewal purposes, the government will provide an interest subsidy of 1.5 percent on the principal of the fixed-asset loans associated with these equipment renewal projects. (7) The central government additionally directs where data centers should be built by lowering administrative review times in locations chosen to maximize utilization and localized service. (8)

State Owned Enterprises (SOEs) also play an important role in driving investment to AI infrastructure. During a meeting held by the State-owned Assets Supervision and Administration Commission (SASAC) of the central government in February 2026, central SOEs were called to actively expand effective investment in computing power and advance the coordinated development of “computing power + electric power”. (9) In January 2025, the Ministry of Industry and Information Technology (MIIT) and the Ministry of Finance jointly established the National AI Industry Investment Fund (the “AI Fund”). (10) With a total capital of ¥60 billion ($8.2 billion) raised from SOEs, the AI Fund mainly invested in AI chips, computing infrastructure, and core integrated compute components, exemplified by its January 2026 investment on Xheart, a company specializing in integrated circuit chip design and the R&D of autonomous driving technologies. (11,12)

Some similarities can be found with recent developments in US industrial policy. In 2025, the White House put out an Executive Order aiming to facilitate permitting for data centers, primarily over-riding previous environmental protection regulations; it differs from similar Chinese efforts however in that it does not require developers to follow any specific strategic directives. (13) In terms of direct investment, the 2022 CHIPS Act included awards in the form of grants and subsidies to support US semiconductor manufacturing, including $8 billion in direct funding to Intel. (14) The US government subsequently required taking a 10% ownership stake of Intel as a condition for having the remaining two thirds of the grant paid out, (15) following a model closer to Chinese SOEs, although this development was recently contested in court by other shareholders. (16) Funding from the US government is more commonly transferred to the technology companies building data centers through contracts for services, with around $7 billion in yearly cloud contracts estimated for FY 2022. (17)

B. Making Data Available for AI

1. Domestic Data Elicitation and Flows

(a) Making Public Data Available for AI

Recent Chinese policy developments increasingly recognize the acquisition and processing of extensive datasets as a fundamental competitive asset in Artificial Intelligence. Data has been officially designated as the “fifth factor of production” by the State Council since at least 2020 (18), particularly public data. However, making this data more available to AI systems has required several systematic overhauls.

The regulatory framework for data use and custody by public institutions has evolved through distinct phases. In 2022, the “Data Twenty Measures” (19) introduced a “Three-Rights Separation” framework to address the complexity of applying property rights to data. By introducing a distinction between data resource ownership, processing and usage rights, and data product disposal rights, the policy moved from implicitly treating data property as an indivisible bundle into a more modular regime. In particular, the right to Process and Use Data enables processors to generate new commercial value by leveraging data without compromising its status as a public asset. The September 2024 Opinion on Accelerating the Development and Utilization of Public Data Resources (20) solidified this Authorized Operation model. The policy also started a shift from mandating institutions to enact “open data” policies for the sake of transparency toward a “data development” requirement directed at making the data economically valuable, authorizing specialized entities to manage and monetize public datasets under state supervision. In early 2025, the central government further scoped the commercial mechanisms for using data in such a way, directing institutions to assess a fair fee to dispose of the data for commercial use (public benefit use remains free) as well as invest in making procedures easier to follow. (21)

The responsibilities for implementing these procedures are shared between the central and local governments. The National Public Data Resource Registration Platform was launched for trial operations on 1 March 2025 to establish a unified national public data circulation ledger and formulate an evaluation logic for data transactions, with the goal of enhancing public data resource management and sharing across organizations. (22) This ledger connects data sources curated and managed by local governments and standardizes technical access requirements under a single user account. The local governments in turn have autonomy to decide what data to prioritize sharing, determine fee structures under centrally determined maxima, and make other context-dependent technical decisions. (23) Cities like Wuhan (24) and Guangzhou (25) for example have published their detailed guidance on public data resource transactions.

Data access for AI in the US follows a different model — combining federally supported research infrastructure, collaboration agreements with leading technology firms, and private licensing markets — rather than the nationally administered commercialization of public institutional data described above. For the largest commercial developers, training data is often already plentiful: vertically integrated cloud and platform operators that build models also control large proprietary collections of user-generated, search, and commerce data, and partnership arrangements among those incumbents can exchange additional inputs unavailable to smaller actors. (26) The National AI Research Resource (NAIRR), established in January 2024 under Executive Order 14110 and transitioning toward a sustained national infrastructure, addresses some of these barriers in the context of scientific research by connecting U.S. researchers and educators to shared compute, models, and datasets through competitive allocations and industry-contributed resources. (27) Additionally, the NSF’s Public Access Plan 2.0 and required Data Management and Sharing Plans extend open-science obligations to scientific data produced under federal research awards — mobilizing federally funded research outputs rather than the institutionally held public datasets China is authorizing for commercial use. (28) The US also aims to direct data to model developers through public-private partnerships. The Genesis Mission reflects a comparable effort to mobilize federal scientific data for AI, but channels access through agreements with a small set of established partners rather than through open registration; (29) whether that structure will extend beyond those incumbents is not yet specified in the mission’s founding instruments. (30) Beyond those sources, supplemental training data is secured through private licensing deals whose reported values run from millions of dollars per year to quarter-billion-dollar multi-year agreements (31), or through copyright litigation — as in suits by news publishers against major model providers (32) — with either route favoring developers that can bear those costs or litigate them to resolution.

(b) Easing Compliance with Data Security Laws

Section I outlines the Three Paths framework. The 2024 Provisions on Promoting and Standardizing Cross-Border Data Flow (the “New Regulations”) (33) are the instrument that has most directly recalibrated its application for commercial and AI-related cross-border flows, moving from a rigid “security-first” posture to a stated goal of balancing development and security requirements, through mechanisms such as raising thresholds for mandatory security assessments and introducing broad exemptions for commercial activities deemed routine. For most AI companies who are non-Critical Information Infrastructure Operators (non-CIIOs), the New Regulations increase the exemption threshold for non-sensitive PI tenfold, from 10,000 to 100,000 individuals (34) processed annually. This relaxation applies strictly to non-Sensitive Personal Information (non-SPI), whereas Sensitive Personal Information (SPI) remains under strict scrutiny.

The exemption regime builds on China’s Data Classification and Grading System, introduced in the Data Security Law. (35) Under this system, the New Regulations establish that data containing neither Personal Information (PI) nor “Important Data” is generally free to flow. While the definition of important data remains a source of ambiguity, the centralized government has produced further overall guidance (36) and tasked local governments and industry regulators (37) with maintaining specific “Important Data Catalogues.” Free Trade Zones (FTZs) are also experimenting with even more limited application of the three paths requirements, by restricting them to “Negative Lists” of even more specific data types.

A key additional provision in the New Regulations is the Data Transit exemption, (38) whereby data collected outside of China, that does not include Personal Information of Chinese citizens or “Important Data” under the grading system, and that is intended for processing and re-export does not trigger “Three Paths” requirements. This allows Multi National Corporations (MNCs) to move their data processing operations to China’s compute infrastructure with greatly facilitated compliance with China’s data security laws. Furthermore, the regulations include exemptions for transfers of PI classified as “truly necessary,” such as contractual necessity (e.g., cross-border shopping, remittances, travel bookings, and visa processing), HR management (facilitating unified global human resource practices under established labor rules), and emergency situations.

2. Importing Data, Exporting Compute and Infrastructure

China’s global strategy also prioritizes developments that will get the country’s industry access to more international data and use cases. The Global AI Governance Action Plan (39) and the Global Cross-border Data Flow Cooperation Initiative (40) showcase a focus on promoting cross-border data flows, and on building shared data platforms and global digital infrastructure: all treating international data as a critical input to AI development. The Global AI Governance Action Plan in particular calls for the construction and sharing of digital infrastructure (including data centers and computing systems) and cooperation in AI applications and the exchange of best practices to standardize data utilization; extending the reach of China’s AI systems and data pipelines.

Investments in global digital infrastructure constitute the base layer of this approach. A central example is the Digital Silk Road (DSR), a component of the Belt and Road Initiative. This state-led framework focuses on digital and technological infrastructure including telecommunications networks, fiber-optic cables, data centers, cloud services, and smart city systems. It is financed by state-backed institutions such as the China Development Bank and the Export-Import Bank of China and large Chinese technology companies such as Huawei and Alibaba. (41) The Digital Silk Road improves digital connectivity across regions such as Asia, Africa, and the Middle East, thereby expanding the global presence of Chinese technology companies and increasing China’s role in global data storage, transmission, and processing networks; as well as acting as a major lever of political influence. (42) Work with state organizations to contract Chinese firms for their infrastructure projects creates long-term technological relationships. (43) For example, Huawei has built telecommunications networks and data infrastructure across multiple African countries, where Western firms have been less active in large-scale infrastructure deployment. In 2024, Huawei partnered with MTN and China Telecom to expand 5G, cloud, and AI capabilities across the continent. (44) These approaches contrast with those of US-based initiatives like Meta’s now-defunct Connectivity program, which aimed to expand internet access in hardest-to-reach places through experimental technologies such as drones and satellite systems, (45) or SpaceX’s Starlink, which focuses on developing satellite-based connectivity through a private business model rather than state-related infrastructure investment. (46)

The international deployment of AI technologies specifically is a key component of this strategy, as reflected in the Global AI Governance Action Plan. In July 2025, China proposed the establishment of a new global AI cooperation organization, along with expressing offers to share its development experience and products with other countries, particularly the “Global South”. (47,48) These efforts are aligned with the document’s framing of AI as a “global public good” and emphasis on sharing technological achievements internationally. (39) Chinese companies are starting to act on this mandate. Alibaba’s cloud division, for example, has expanded its global infrastructure by launching a second data center in Dubai in 2025, as part of a broader plan to invest heavily in international AI and digital infrastructure. (49) Similarly, Chinese AI startup Zhipu AI has accelerated its overseas expansion by collaborating with Alibaba Cloud and establishing operations in regions such as Southeast Asia and the Middle East, while promoting localized AI applications for governments and enterprises. (50) In addition, Chinese autonomous driving companies, such as WeRide, have expanded into the Middle East market. They have partnered with local governments in cities such as Dubai and Abu Dhabi to test and deploy robotaxi services, with plans for large-scale commercial operations in the coming years. (51) These deployments of AI systems abroad generate new data and operational experience, which may generate operational data and use-case experience as inputs to domestic model development, not only expanding influence abroad but also supporting the strategy’s goal of access to international data and applications.

Bipartite diplomatic relationships beyond direct exchange of commercial services also play a major role in the overall strategy. The China–EU High-Level Digital Dialogue (52) serves as a regular government-to-government platform for coordination on digital policy — including AI regulation, platform governance, and cross-border data flows — and has recently facilitated practical cooperation on industrial data transfer, including a mechanism to streamline the transfer of non-personal industrial data in sectors such as finance, automotive, and information technology. (53)

C. Self-Reliance through Open-Source AI and Domestic Chips

Sustainable success for the Chinese AI industrial strategy also requires tailoring its development and deployment to the country’s full technical stack and ecosystem. The necessity of this push for self-reliance has been reinforced in recent years by U.S. restrictions on advanced semiconductors exports, (54) requiring Chinese technological companies to both ramp up the development of domestic chips and to prioritize computational efficiency in the pursuit of more performant models. The development of an open source and open weight AI ecosystem has proven particularly effective to that end, allowing different model developers to benefit from each other’s technical contributions to accelerate overall technology development. (55)

To promote such an ecosystem, initiatives such as the* International Open-Source AI Cooperation Initiative* (56)* *emphasize both lowering barriers to open AI development domestically and encouraging global participation. This emphasis on openness is backed by concrete policy incentives operating at multiple levels. At the national level, the central government has launched top-down initiatives to embed AI into education, industrial planning, and public governance. Local governments have developed their own instruments, including compute vouchers, tax concessions, and discounted data center access, to enable smaller developers to train or adapt their own AI models, making distributed development viable even for actors without large capital reserves. (57) These instruments have successfully supported the development of more compute-efficient approaches to training and deploying performant models, with the popularity of DeepSeek’s R1 and subsequent models standing as a notable example. (58) The focus on efficiency has adapted to the more limited availability of compute resources and enables the adoption of Chinese-trained models abroad, furthering the adoption of Chinese-designed standards for models and data formats and enabling further innovations built on these models to flow back to their developers.

On the hardware front, commercial restrictions on the availability of top-of-the-line NVIDIA chips have also accelerated domestic efforts in China to develop alternative AI hardware and optimize model performance under limited compute conditions. Chinese technology firms, including Huawei, have advanced the development of domestic AI chips such as the Ascend series, which are increasingly used for AI systems within China. (59) These chips have become an important component of China’s AI infrastructure, particularly in state-supported and enterprise applications. Meanwhile, model developers have worked on ensuring that their AI systems run as efficiently as possible on domestically developed chips. Zhipu AI highlighted efforts to make the GLM 5 model inference run efficiently on a wide range of domestic chips at the time of release. (60) More recently, Meituan also announced it had trained a trillion-parameter model entirely on domestic chips, (61) and OpenBMB, a partnership with Tsinghua University, released an ultra-efficient open language model for edge devices trained on Huawei Ascend chips. (62)

The integrated approach of providing mandates and incentives for open-source development, domestic chip manufacturing, and mutual integration of the two has allowed Chinese developers to play to the strengths of their comparatively more distributed technological ecosystem; rather than simply attempting to catch up in the footsteps of their US counterparts with more limited resources. This approach is starting to bear fruit in the development of “token factories” serving these models with certified compliance and fully domestic infrastructure; such as the Shantou Free Trade Zone, (63) showing a different model of generative AI use at scale than the US market dominated by a few commercial providers, with more modularity across stack layers and more points of intervention for compliance.

III. Addressing New Market Power Factors and Data Risks of AI

1. Adapting Platform and Competition Law to AI

(a) Updates to the Anti-Monopoly and the Anti-Unfair Competition Laws

The development of AI technologies is inherently tied with the growing importance of a platform-based economy. Platforms use technologies such as dynamic pricing, demand forecasting, advertising optimization, content recognition and moderation, and algorithmic recommendation to continuously optimize the supply of their goods and services. They are also a particularly attractive support for the deployment of generative AI based systems, given their broad user bases and the prevalence of digital interactions that are immediately legible to the system. This makes them a key driver of AI development and deployment, particularly in the Chinese context where platforms mediate an even greater share of online interactions. (1) Consequently, the successful development of China’s AI ambitions is intrinsically linked to the strategic cultivation of its platform economy. This drive is reflected in the 2025 Outline of the Government Work Report, (2) which requires the added value of “core digital economy” industries to account for 12.5% of the GDP, a significantly higher target than previous years. This national directive is reinforced by local government initiatives, such as the Sichuan provincial government’s annual fiscal support for the platform economy to the amount of 200 million RMB yearly for the next three years. (3)

At the same time, competition in the AI and platform economy has evolved from traditional price wars into comprehensive competition with data, algorithms, and capital as increasingly major factors of market power. This shift is evident in market dynamics, such as the impact of Alibaba’s ability to leverage its capital advantages over competitor MeiTuan, and the rapid expansion of platforms like TikTok (a short video platform owned by ByteDance) and Rednotes (an online community to share life experiences) supported by new modes of interaction between e-commerce and personal digital lives. This evolution has necessitated recent amendments to the Anti-Monopoly Law and the Anti-Unfair Competition Law.

The 2022 revision of the Anti-Monopoly Law (AML) (4) adapted the existing framework to the platform economy. The legislature incorporated concepts such as data, algorithms, technology, and capital advantages, signaling that digital market competition requires tools beyond traditional industrial-era metrics. The AML updated provisions – including on hub-and-spoke agreements, abuse of dominance through data and algorithms, and below-threshold merger review – to recognize that control over data, user lock-in, and platform ecosystems, rather than turnover alone, determines competitive significance; particularly in an ecosystem where digital control may matter more than short-term turnover and price impacts. Furthermore, the inclusion of civil public interest litigation provides a collective mechanism for addressing the dispersed, low-value harms that individual users often face when their rights are infringed by platforms. (5) These efforts continue a trend started with the 2021 Anti-Monopoly Guidelines for the Platform Economy Sector (6) addressing the role of technical advantages and setting foundations for addressing specific concerns about how recent generative AI systems raise the risk profiles of new kinds of personal data. (7)

The 2025 revision of the Anti-Unfair Competition Law (AUCL) (8) complements the AML by focusing on unfair competitive conduct within digital platforms. While the AML addresses market structure and dominant power, the AUCL targets specific on-platform dynamics including digital fraud and misrepresentation of goods, imposition of unfair payment prices and conditions, and lock-in through technical means. This focus reflects a recognition that the same AI-enabled tools that improve efficiency can also facilitate exclusion, manipulation, and unfair advantage. It places greater compliance responsibilities on platform operators and strengthens penalties, addressing how large digital intermediaries may distort fair competition without necessarily fitting older monopolization models. Regarding AI specifically, the regulation explicitly prohibits training products on data scraped from the platforms, as well as using AI to distort markets by fabricating products or traffic. The 2025 AUCL also includes an extraterritorial jurisdiction clause (Article 40), (9) as a critical tool to enforce the use of approved channel for training models, both domestically and abroad.

Among other goals, these legal reforms can be read as a way to encourage more stability of the market structure and dynamics between different actors: the overall approach allows and even encourages a greater role of AI technology and data use in improving the economic performance of the digital sector, but also acknowledges ways in which the technical paradigm might disrupt the status quo on which the overall industrial strategy is built; a status quo in which no single private actor should have complete ownership of any given sector.

(b) Comparison of US and Chinese Cases

Both China and the United States have begun to recognize that the use of data, algorithms, and platform technologies may create new competition law problems in the digital economy. China’s approach is primarily rule-based, responding through statutory amendments and judicial interpretations (e.g., the 2022 AML revision and 2024 Judicial Interpretation of the Supreme People’s Court on monopoly civil disputes (10)). In contrast, the United States relies more heavily on case-driven enforcement actions and litigation under existing antitrust statutes (e.g., the Sherman Act and the FTC Act).

The issue of algorithmic collusion showcases these dynamics, with a different balance between *ex ante * rule-making and litigation. According to the 2024 Judicial Interpretation of the Supreme People’s Court, where undertakings use data, algorithms, technology, or platform rules to exchange information, coordinate conduct, or achieve concerted behavior, the conduct may be reviewed under the rules on monopoly agreements. This shows a growing recognition of the specific risks of algorithmic collusion, although there are still relatively few mature public cases in which AI algorithms themselves are the core issue. In the United States, this issue was addressed in the U.S. v. RealPage case, (11) in which the DOJ argued that algorithmic pricing tools may facilitate coordination among market participants. The case ended in a settlement that included prohibition on different forms of data use for training and deploying AI models, particularly non-public data and data from competitor platforms – similar to provisions in the revised AUCL. The settlement also supported private class actions (12) by affected individuals seeking financial remedy.

Issues related to self-preferencing and access restrictions show similar dynamics. According to Articles 9 and 22 of China’s 2022 Anti-Monopoly Law, undertakings may not use data, algorithms, technology, capital advantages, or platform rules to engage in monopolistic conduct, and dominant undertakings may not use such means to abuse market dominance. These provisions provide a legal basis for addressing practices such as self-preferencing, ranking manipulation, or access restrictions; but in practice they have been operationalized through administrative decisions more than judicial judgments. In Qihoo 360 v. Tencent, Tencent blocked competing applications and restricted interoperability; Qihoo sued for abuse of dominance. (13) The court held that multi-homing costs are low and alternatives exist, and blocking does not automatically foreclose competition; and in general tends to reject a dominance finding based only on a lack of technical interoperability. Administrative decisions, on the other hand, have taken a stricter stance, such as by mandating Alibaba to open its data, payment system, and application access. (14) Conversely, the same question in the US has been addressed chiefly through litigation. In FTC v. Amazon (15), Yelp v. Google (16), and the Department of Justice’s litigation against Google (17), U.S. authorities have been more willing to frame platform self-preferencing and distribution restrictions as concrete antitrust issues, although the remedies have been behavioral rather than structural and limited in scope. (18)

Finally, efforts to address data-driven price discrimination in both China and the United States both show the limitation of antitrust enforcement to address this concern. According to the Supreme People’s Court’s 2025 discussion of platform monopoly issues, (19) “big data discrimination” is recognized as a new category of competition concern in the platform economy. However, in China, this issue still appears more often in policy discussions and legal interpretations (20) than in mature antitrust judgments. In judicial practice, courts have not upheld claims concerning “big data price discrimination against loyal customers”. Claims have been rejected on grounds such as the user’s prior consent to data use, the view that price fluctuations were caused by factors other than data, the user’s ability to switch platforms, and insufficient technical evidence. (21) In the United States, according to the FTC’s 2024 surveillance pricing inquiry, regulators have also begun to examine the use of consumer data, behavioral information, and algorithmic tools in personalized pricing. However, in the absence of the direct use of competitors’ data (as in the RealPage case outlined above), the courts have tended to favor deployers of algorithmic pricing software. (22)

2. Disclosure Requirements and Content Rules for AI Technology

AI models developed in China are subject to substantial transparency requirements, which apply to both proprietary and open models. The requirements are implemented through an algorithmic filing system, which requires developers to provide information about the system’s training data, development conditions and main technical characteristics, and testing. (23)

The algorithm filing system, officially known as the Internet Information Service Algorithm Filing System, was established primarily under three key regulations: Algorithmic Recommendation Provisions (2022), (24)* Deep Synthesis Provisions (2023),* (25)* Generative AI Measures (2023)* (26)*. *Managed by the Cyberspace Administration of China (CAC), filing is mandatory for any entity providing services in China that uses algorithms with “public opinion attributes” or “social mobilization capabilities.” This essentially covers content platforms, e-commerce & service apps, AI developers, and search engines. Centered on the function of disclosure, the algorithm filing system enables Chinese regulators easier access to instrumental information to support rule-making about technical systems.

The algorithm filing focuses on AI service providers’ behavioral logic. Pursuant to the Provisions on the Administration of Algorithm Recommendations for Internet Information Services (24) and the Provisions on the Administration of Deep Synthesis of Internet Information Services (25), algorithm filing should provide information to help regulate algorithmic behavior, protect user rights, and safeguard public interests while preventing abuses such as algorithmic discrimination, information cocoons (more often referred to as echo chambers or algorithmic bubbles in the US context), and the generation of “false information” or non-compliant content. Service providers of algorithm recommendations, deep synthesis (including technical supporters), (24) and generative AI that possess public opinion attributes or social mobilization capabilities (internet information services such as forums, blogs, chat rooms that provide channels for the expression of public opinion or possess the capability to mobilize the public to engage in specific activities) must complete the “Internet Information Service Algorithm Filing.” The generative AI services that have completed filing include DeepSeek and Baidu’s Ernie Bot, as shown on the CAC’s website. (27) According to the system guidelines, filing categories include generation and synthesis, personalized recommendation, ranking and selection, retrieval and filtering, scheduling and decision-making. AI enterprises must display their filing number in a prominent location on their service platforms, provide a link to the public information, and disclose the basic principles of their algorithmic mechanisms (24).

Large language models fall within this filing system; where they are deployed as generative AI services with public opinion attributes or social mobilization capabilities, the Interim Measures for the Management of Generative Artificial Intelligence Services (2023) (26) require more granular disclosure and assessment, enforced by the CAC in coordination with MIIT and MPS. The LLM filing process is bifurcated into technical registration and security assessment. First, developers must submit detailed disclosures to the CAC’s filing portal. This includes the model’s parameter scale, architecture, and the specific “core logic” used for information retrieval and generation. Second, a critical component of the filing is the disclosure of training data sources. Developers must demonstrate the legality of their datasets, including methods for filtering “illegal or harmful” information and the specific ratios of domestic to international data. Third, Chinese law requires a mandatory “Security Self-Assessment.” This involves testing the model against a standardized corpus of thousands of sensitive prompts to ensure outputs align with “Socialist Core Values” and do not threaten national security or social stability.

The Chinese regulatory scope encompasses both proprietary “Black Box” models (e.g., Baidu’s Ernie Bot) and open-weight models (e.g., Alibaba’s Qwen, DeepSeek, and GLM). For open-weight developers, the filing serves as a “licensing” mechanism; even if the weights are distributed globally, the developer remains legally responsible for the “base safety” of the model within Chinese jurisdiction. Overall, in accordance with the Interim Measures for the Management of Generative AI Services and the Provisions on the Security Assessment of Internet Information Services with Public Opinion Attributes or Social Mobilization Capabilities, (28) the primary goal of LLM filing is to regulate the launch of generative AI services.

Conclusion

China’s AI policy operates as a coordinated industrial strategy in which the state directs investment, sets adoption targets, and adapts pre-existing digital-governance tools to new technical paradigms ex ante – in contrast to a US approach that more often shapes AI-related constraints through litigation outcomes under statutes not designed for the technology, and that responds to the priorities of a small number of concentrated private developers in terms of directing investment in supporting resources. Binding instruments such as the AI+ Strategy Guidance and the 15th Five-Year Plan, together with a centralized legislative framework and SOE-backed capital, give the central government sustained leverage over infrastructure, research priorities, and sectoral uptake, with state-owned enterprises filling gaps where private investment is slow or returns are uncertain.

Supply-side policy concentrates on computing power, energy, domestically viable chips, and data as a competitive asset – areas where China has scaled physical infrastructure through fiscal capital and SOE investment at a pace that reflects strategic priority rather than market timing alone. On data, recent reforms work within – and extend – the privacy and cybersecurity frameworks established under the Personal Information Protection Law, the Data Security Law, and classification-and-grading rules, recalibrating them toward a new state presented as a balance of development and security rather than replacing them: thresholds for compliance triggers have been raised and broad exemptions introduced for routine commercial flows and data transit, while sensitive personal information, “important data,” and high-risk categories remain subject to stricter scrutiny. Access to training data follows a similarly distinct logic. Where US AI development has been shaped heavily by copyright disputes and private licensing arrangements — supplementing the proprietary platform data held by vertically integrated incumbents — that favor firms with the capital to negotiate exclusive access, China has moved public datasets toward authorized commercial use under state supervision — through mechanisms such as three-rights separation, a national registration platform, and locally administered fee structures – treating data held by public institutions as a factor of production to be developed for economic value rather than as a transparency-only resource. International initiatives, from the Global Cross-Border Data Flow Cooperation Initiative to Digital Silk Road investments, extend this logic by importing data and use cases while exporting compute and AI services. U.S. export controls on advanced semiconductors have reinforced a parallel push toward open-source development and domestic hardware, producing a more modular ecosystem than the US market dominated by a few closed providers.

Finally, competition-law reforms and AI-specific governance rules address the market-power and social risks that scale deployment introduces – not only to protect users, but to preserve the market structure and internet content controls on which the broader industrial strategy depends: a digital sector in which AI and data use may drive growth, but no single private actor comes to exclusively control a sector or the channels through which models are trained and deployed. Revisions to the Anti-Monopoly Law and the Anti-Unfair Competition Law treat data, algorithms, and platform ecosystems as sources of market power accordingly, whereas the United States addresses overlapping concerns chiefly through litigation under existing antitrust statutes, with remedies that have tended to remain behavioral and limited in scope. Mandatory algorithm and generative-AI filing adds a further layer: developers must disclose training data, technical characteristics, and pass security self-assessments before launch – a process-oriented regime without a direct US equivalent to date.

Together, these elements define a Chinese model that is at least as committed to the rapid expansion and integration of AI technology as the US, but shows structural differences in terms of who holds leverage over development and when regulation intervenes: upstream process requirements and competition rules distribute control across the stack under state oversight, whereas the US more often cedes direction to concentrated private developers and addresses the same risks through ex post litigation once market structures have formed.