Report on Promoting Artificial Intelligence Cooperation between

Report on Promoting Artificial Intelligence Cooperation between Guangxi and ASEAN with Semantic Sovereignty as the Core

Yucong Duan

International Standardization Committee of Networked DIKWPfor Artificial Intelligence Evaluation(DIKWP-SC)

World Academy for Artificial Consciousness(WAAC)

World Artificial Consciousness CIC(WAC)

World Conference on Artificial Consciousness(WCAC)

(Email: duanyucong@hotmail.com)

Introduction

Artificial Intelligence (AI) technology is developing rapidly and is profoundly changing the global economic and social landscape. Against the backdrop of accelerated digital transformation in ASEAN countries and China's proposal to build a community with a shared future for mankind, Guangxi Zhuang Autonomous Region, relying on its geographical and cultural advantages, is actively exploring new paths for open AI cooperation facing ASEAN. The Guangxi Artificial Intelligence University was recently officially established. With the positioning of "Based in Guangxi, Facing ASEAN," it aims to build a platform integrating talent cultivation, technology R&D, industrial incubation, and international cooperation. The college has attracted 42 high-level scientific research teams and 330 master's and doctoral students, dedicated to serving the high-quality development of Guangxi's AI industry and assisting in the construction of the China-ASEAN Artificial Intelligence Innovation Cooperation Center. Currently, under the guidance of the national new generation AI strategic deployment, we have the conditions and necessity to build Guangxi into a highland for AI cooperation facing ASEAN, realizing trustworthy and reliable multilingual communication and cross-border flow and service of data, information, and knowledge among countries.

However, opportunities come with challenges. On one hand, ASEAN countries have shortcomings in AI technology, capital, and talent, and have an urgent desire to strengthen cooperation with China. China and ASEAN are highly complementary in the AI field. On the other hand, issues such as semantic opacity and "semantic bias" in the cross-language and cross-cultural application of AI systems are becoming increasingly prominent, posing new challenges to cultural security and international discourse power. In particular, the rise of large language models makes it possible for the languages and cultures of different countries to be distorted by the biases of external algorithms. If countries cannot master autonomous semantic technologies and AI models, it will be difficult for them to have a voice in the global semantic system. Based on this, Chinese scholar Professor Yucong Duan took the lead in proposing the concept of "Semantic Sovereignty," advocating enhancing national autonomy at the semantic level of AI to ensure that the country's language, culture, and values are accurately and fairly expressed in the digital age. Semantic sovereignty is consistent with the concept of "Sovereign AI" discussed internationally in recent years: Sovereign AI refers to a comprehensive AI system independently developed, trained, and controlled by a country, reflecting national will and value extension. Jensen Huang, founder of NVIDIA, emphasized at the 2024 World Government Summit that countries need to own their own AI infrastructure to protect cultural and economic security. Professor Yucong Duan further pointed out that Sovereign AI should possess advanced learning, perception, reasoning, and self-cognitive capabilities, be able to make decisions and interact within a specific cultural context, and reflect domestic cultural, social, and ethical values. Therefore, semantic sovereignty and sovereign AI are two sides of the same coin: the former focuses on autonomy at the content and meaning level, while the latter focuses on autonomy at the algorithm and computing power level. The combination of the two builds the cornerstone of national sovereignty and cultural security in the digital age.

Central leadership attaches great importance to sovereignty and security issues in AI development. General Secretary Xi Jinping has repeatedly emphasized the need to "accelerate the establishment of a new generation AI key generic technology system, grasp the layout on shortcomings, and ensure that key core technologies of AI are firmly held in our own hands." "Semantic Sovereignty" is precisely a new topic of digital sovereignty in the AI era. It elevates the focus from the control of data itself to the control of the meaning carried by data. This is of great significance for maintaining national cultural security and enhancing international discourse power. To implement this concept, we need to take the lead in practicing the principle of "Semantic Sovereignty" in Guangxi's AI cooperation facing ASEAN. Through technological innovation and institutional design, we must ensure that cross-language AI systems are trustworthy and reliable, disseminate China's mainstream values, and guarantee the safe and controllable cross-border flow of data and knowledge. Based on the above background, a report on deepening Guangxi-ASEAN AI cooperation and implementing the concepts of Semantic Sovereignty and Sovereign AI is hereby presented.

Background and Situation

(I) Global AI Competition and the Rise of "Sovereign AI"

Currently, AI has become a strategic highland for a new round of global technological revolution and industrial transformation. Countries are racing to invest in AI R&D, widely applying AI in medical care, education, transportation, finance, and other fields to improve efficiency and spawn innovation. At the same time, AI is also seen as a new carrier of national comprehensive strength and security. Building a "Sovereign AI" system with national characteristics has risen to be an important part of national science and technology strategies. Jensen Huang proposed that "every country needs to own its own AI infrastructure" to protect its own culture and economic security. This view reflects the universal concern of the international community for Digital Sovereignty. From data sovereignty and cyber sovereignty to today's algorithmic sovereignty and semantic sovereignty, countries are attempting to build autonomous and controllable AI systems to adapt to and lead technological changes.

However, China also faces a series of shortcomings and challenges in the process of building "Sovereign AI": First, the supply and demand of computing power are unbalanced, high-end chips are controlled by others, and there is still a gap between ultra-large-scale computing power infrastructure and international advanced levels; Second, large models are homogeneous, lacking unified planning and differentiated innovation, and the underlying architecture of many domestic large models still relies on foreign open-source frameworks; Third, emphasis on application over foundation, original basic technologies and core algorithms are weak, and the phenomenon of "top-heavy" is prominent; Fourth, high-quality Chinese corpus is scarce, English accounts for over 90% of global AI training data, while Chinese accounts for less than one-thousandth, local models are restricted in corpus and knowledge, which is unfavorable for reflecting China's values; Fifth, basic talent is in short supply, top AI talents, especially basic research talents, are scarce, and brain drain is serious. These problems indicate that building sovereign AI is both an opportunity and an arduous task, and must coordinate the "dual-wheel drive" of technological breakthroughs and governance innovation.

(II) Semantic Risks Highlighted and the Proposal of Semantic Sovereignty

In AI governance, the widespread application of Generative AI represented by large language models has triggered a series of systemic risks at the semantic level. Prominently manifested as: "Semantic Black Box"—the AI decision-making process lacks explainability, the internal semantic reasoning link is invisible to humans, outputting conclusions without explaining meanings, making it difficult for regulators and the public to judge its rationality and credibility; "Semantic Bias"—model training inherits or even amplifies historical biases implicit in the corpus, which may produce unfair tendencies towards certain groups or concepts, and the semantic systems of different cultures may be intruded upon and distorted by the biases of other countries' models; "Value Misreading"—the same semantics have different connotations in different cultures, current AI lacks sensitivity and may misjudge or misread values in cross-cultural contexts, leading to AI outputs violating local ethical norms or even causing cultural misunderstandings. For example, models trained in the West may produce biased understandings when answering traditional Eastern ethical questions due to a lack of corresponding corpus and value labeling. Multilingual large models that "dismantle language and value barriers" promote globalization but also pose new challenges to national sovereignty and cultural security—how to avoid AI outputting content that violates China's mainstream values and incorrectly interprets policies and laws has become an urgent problem to be solved in AI governance.

It is in this context that the concept of "Semantic Sovereignty" emerged. It was first proposed by Professor Yucong Duan of Hainan University in 2023, referring to the right and ability of a country or region to have its own language, culture, values, and knowledge system accurately, completely, and fairly expressed in the global digital environment. This concept is a strategic upgrade of traditional "data sovereignty"—rising from focusing on data itself to focusing on the meaning and value behind data. Simply put, data sovereignty focuses on the control of "data," while semantic sovereignty further emphasizes the control of the "meaning carried by data." Semantic sovereignty embodies the cross-fusion of the human rights dimension of language rights, the value dimension of cultural sovereignty, and the technical dimension of digital sovereignty, directly responding to the core question of "who has the right to define and interpret meaning in the digital space." At the level of national security and digital governance, the importance of semantic sovereignty is no less than that of data sovereignty: Data Sovereignty ensures the state's jurisdiction over data resources, while Semantic Sovereignty ensures the state has the ultimate decision-making power over the meanings and value orientations understood and output by AI. By guaranteeing semantic sovereignty, we can prevent our national language and culture from being distorted by external algorithms, so that our language and culture will not be submerged in the biases of other countries' models.

(III) Opportunities for China-ASEAN Digital Cooperation and Guangxi's Mission

The ASEAN region is in a critical period of digital transformation, and countries have urgent and diverse needs for advanced AI technologies. However, ASEAN generally faces bottlenecks such as shortages of technology, funds, and talent, and urgently needs external cooperation for assistance. As a major country in AI technology and application, China is highly compatible with ASEAN's development stage and needs—China's booming AI technology can meet ASEAN's diversified needs. ASEAN is willing to strengthen digital economy cooperation with China and build a closer partnership for scientific and technological innovation. In 2023, the "China-ASEAN Information Harbor" was upgraded, and the first "China-ASEAN AI Ministerial Roundtable" was held in 2024, where all parties reached a consensus on consolidating the foundation for AI development, providing open-source and open services, strengthening industrial cooperation and docking, and promoting talent cultivation. ASEAN Secretary-General Kao Kim Hourn stated that China's huge digital ecosystem and leading platform technologies will provide strong support for cooperation between the two sides. Both sides should grasp global trends such as AI, robotics, IoT, and cloud services, jointly cultivate unicorn companies and tech giants, and build the region into an innovation highland. At the same time, the upgrade agreement of the "China-ASEAN Free Trade Area 3.0" is about to be signed, and the Digital Economy is listed as the first of the new chapters. Through rule alignment, standard mutual recognition, and mechanism interconnection, the new agreement will facilitate the smooth landing of AI solutions in the ASEAN market and promote greater effects of digital economy cooperation between the two sides.

As the forefront and window of China's opening up and cooperation with ASEAN, Guangxi is at the junction of the two major markets of the whole country and ASEAN, with geographical proximity and cultural affinity, possessing unique advantages. General Secretary Xi Jinping endowed Guangxi with the new mission of "Three Positions," including building an important gateway for the organic connection of the "Belt and Road." In recent years, Guangxi has focused on "what the country needs, what Guangxi can do, and what ASEAN expects," fully promoting strategic docking, policy exchange, and application cooperation in AI with ASEAN countries. Guangxi has proposed and practiced the characteristic development path of "R&D in Beijing-Shanghai-Guangzhou + Integration in Guangxi + Application in ASEAN," that is, docking with the cutting-edge AI technologies of first-tier cities like Beijing, Shanghai, Guangzhou, and Shenzhen, conducting integration and adaptation in Guangxi, and then exporting to the vast ASEAN market. Guangxi is accelerating the construction of the China-ASEAN AI Application Cooperation Center (abbreviated as "Nanning A Center"), striving to create an AI cooperation highland facing ASEAN. This model has significant advantages: Guangxi is rich in minority language talents, and ASEAN corpus resources are abundant, which can be used for AI training; Proximity to ASEAN for fast communication, network latency from Nanning to Vietnam is only at the millisecond level, facilitating data transmission; Easy for integration and transformation, Guangxi can gather domestic mature AI technologies, build China-ASEAN trusted data zones and public service platforms, achieving "bag-in check-in, plug-and-play" services; Low communication costs, geographically connected and culturally similar, significantly reducing cooperation communication costs and shortening project landing cycles. At the same time, Guangxi possesses national-level open platforms such as the China (Guangxi) Pilot Free Trade Zone, China-ASEAN Information Harbor, and Cross-border E-commerce Comprehensive Pilot Zone, piloting in the fields of digital economy and cross-border cooperation, providing policy support for cross-border AI applications. These advantages make Guangxi an excellent place to explore cross-border AI cooperation pilot fields.

Since the beginning of this year, under the guidance of national ministries, Guangxi has frequently made practical moves to layout AI: Established the Autonomous Region AI Development Working Special Class in February, elevating AI development to a prominent position; intensively issued policies such as the "Decision on Accelerating the High-Quality Development of Artificial Intelligence" and the "Action Plan for 'Artificial Intelligence + Manufacturing' (2025-2027)" in the first half of the year, deploying six major actions including AI product breakthroughs, "Intelligence Empowering Ten Thousand Enterprises," key technology breakthroughs, enterprise cultivation, industrial layout optimization, and factor guarantee. It proposed that by 2027, the output value of Guangxi's AI-related industries will exceed 100 billion yuan, creating 100 landmark intelligent products, 100 typical application scenarios, cultivating 10 chain-owner enterprises with national influence, and initially building an AI industry highland facing ASEAN. To stimulate innovation vitality, Guangxi held the "A Super" Super League themed on AI empowering industries. Since its launch in July this year, it has attracted more than 3,300 teams from 31 provinces in China and ten ASEAN countries to participate, emerging a batch of results with landing potential. At the 2025 China-ASEAN AI Cooperation Matchmaking Meeting held in Nanning in September, Guangxi attracted more than 10 billion yuan of AI funds to be signed and landed. 61 domestic AI projects reached cooperation intentions with Guangxi and ASEAN scenarios, of which 54 projects have accelerated landing in Guangxi; at the same time, it facilitated cooperation with 22 overseas enterprises, and 14 overseas enterprises settled in the Nanning A Center, presenting a "magnetic effect," providing strong support for technology integration and market transformation.

In short, under the guidance of national strategy and its own efforts, Guangxi is becoming an important pivot for China-ASEAN AI cooperation. However, as cooperation deepens, we must simultaneously build a trustworthy and reliable cross-cultural AI exchange mechanism. This requires taking the concept of "Semantic Sovereignty" as the command, solving problems of semantic deviation and cognitive distortion in cross-border AI applications through technological and institutional innovation, ensuring that AI exchanges between China and ASEAN countries are explainable, controllable, and trustworthy, providing strong support for the construction of the "Belt and Road" Digital Silk Road and the ASEAN Digital Economy Community.

Core Views and Theoretical Basis

(I) Sovereign AI and Semantic Sovereignty: Integration of Concepts

"Sovereign AI" emphasizes a country's autonomy and control over AI development, ensuring AI conforms to national interests and value systems. It focuses on achieving autonomous control at the algorithm and computing power levels and is the technological carrier of national will in the digital age. "Semantic Sovereignty," on the other hand, focuses on autonomy at the content and meaning levels of AI systems, ensuring that the semantics output by AI meet legal, policy, and mainstream value requirements. The two are not isolated concepts but complement and support each other: Sovereign AI provides an autonomous technology platform, while semantic sovereignty ensures the correct orientation of content on that platform. Simply put, Sovereign AI solves the problem of "who builds and who controls," while Semantic Sovereignty solves the problem of "what to build and what to output." Only when technological sovereignty and semantic sovereignty are realized simultaneously can a country truly control the direction and dividends of AI development.

Professor Yucong Duan's research further clarifies the significance of the integration of the two: Semantic Sovereignty is the "root" of Sovereign AI. Without autonomy at the semantic level, AI technology autonomy can hardly guarantee national cultural security and value transmission; Sovereign AI is the "body" of Semantic Sovereignty. Without autonomous and controllable AI infrastructure, semantic sovereignty will lose its carrier. Therefore, when promoting AI applications, we must not only build autonomous AI infrastructure and large model systems but also ensure that their output content reflects the national context and values, truly achieving both hard technological power and soft cultural power. Guangxi's AI cooperation facing ASEAN should take this as a guiding ideology, integrating the concept of semantic sovereignty throughout the entire process of technology R&D and application output, creating a regional AI cooperation paradigm with Chinese discourse power.

(II) DIKWP Model: Technical Cornerstone of Semantic Sovereignty

To organically integrate Sovereign AI and Semantic Sovereignty, a corresponding theoretical model is needed as a bridge support. Professor Yucong Duan's team originally proposed the DIKWP Model, providing a key technical base for us to realize semantic sovereignty. DIKWP is formed by extending a "Purpose" (Purpose) layer on the basis of the classic DIKW (Data-Information-Knowledge-Wisdom) model, containing five levels: Data (D), Information (I), Knowledge (K), Wisdom (W), and Purpose (P). These five levels constitute a complete semantic cognitive chain, describing the entire process of AI from perception to decision-making. Among them, the newly added "Purpose" layer is particularly critical, making up for the deficiencies of traditional data-driven methods: past AI systems lacked a clear representation of "why use this information," while the DIKWP framework takes the goals and Purposes behind behaviors into consideration. It emphasizes that any transformation of Data $\rightarrow$ Information $\rightarrow$ Knowledge $\rightarrow$ Wisdom should be driven by clear goals and purposes. As research points out: "Without clear goals, it is difficult to complete the closed-loop transformation from data to wisdom." Therefore, in the DIKWP model, the "Purpose" layer covers behavioral goals and motivations, driving and constraining the cognitive process of AI. This allows AI systems to map data, information, knowledge, and wisdom to specific purposes, adopt corresponding strategies for different input types, and achieve efficient and precise decision-making.

The DIKWP model endows the AI system with a "Purpose-driven Data-Wisdom Framework." Through this model, we can comprehensively understand how raw data is transformed into executable wisdom under the guidance of specific goals, thereby ensuring that human value goals and contextual considerations are integrated into the AI cognitive decision-making process. This lays a theoretical foundation for realizing semantic autonomy. In fact, the DIKWP model provides a structured semantic mechanism for semantic governance: at the D layer (Data), it ensures semantic sources are traceable; the I layer ensures consistency of semantic dissemination; the K layer guarantees knowledge reliability; the W layer outputs explainable wisdom; and the P layer locks target Purposes. It can be said that the DIKWP five-dimensional architecture provides a vertically connected "semantic coordinate system," making AI's understanding and processing of human concepts rule-based. Professor Yucong Duan took the lead in applying the DIKWP concept to the practice of semantic sovereignty, developing an AI white-box evaluation system based on the DIKWP model for comprehensive assessment of AI cognitive capabilities, ensuring Sovereign AI conforms to human Purposes. At the 5th China Data Factor 50 Forum in July 2025, he pointed out that to make AI serve humanity better, we must break through the limitations of traditional black-box evaluation and build a white-box evaluation system. The DIKWP-driven white-box evaluation method proposed by the team can comprehensively diagnose and assess the cognitive capabilities of AI systems at the five levels of perception, information, knowledge, wisdom, and Purpose, achieving quantifiable and verifiable capability measurement. Currently, this system has been publicly released, hoping to become an internationally accepted AI capability assessment standard. This practice shows that the DIKWP model not only has theoretical innovation but can also be used to formulate AI governance standards, providing a trustworthy guarantee mechanism for semantic sovereignty.

It is worth emphasizing that the DIKWP model is an important innovation achievement with independent intellectual property rights in China and has successfully entered the international standard stage. In October 2025, the International Organization for Standardization officially released the ISO/IEC TR 25005-2:2025 "Smart City Data Use - Use Case Analysis and Considerations" technical report, which formally included the DIKWP model advocated by the World Association of Artificial Consciousness (WACA). According to the official ISO explanation, the appendix of the standard cites the latest research results of the DIKWP team, using the "Data-Information-Knowledge-Wisdom-Purpose" five-dimensional framework as a reference model for smart city data governance. This is a major breakthrough for the DIKWP method in the field of international standards, marking the entry of the semantic governance concept of Chinese scholars into the vision of global smart city data governance. WACA subsequently established the "International Artificial Intelligence DIKWP Evaluation Standard Committee," led by Professor Yucong Duan, to promote the application of the five-dimensional evaluation framework globally. Professor Yucong Duan stated in a media interview that the adoption of the DIKWP model by international standards is an affirmation of the team's years of research work and will provide new methodological support for the semantic understanding and governance of smart city data. This achievement further proves that the DIKWP model, as the technical cornerstone of semantic sovereignty, has international advanced nature and broad applicability. We should make full use of this independent innovation advantage, integrate the DIKWP model into the technical solutions and standards of China-ASEAN AI cooperation, and establish China's rule dominance in the field of AI semantic governance.

(III) From Concept Space to Semantic Space: Achieving Semantic Alignment

To implement semantic sovereignty in cross-cultural AI cooperation, the problem of Consistency in AI semantic understanding must be solved. Traditional Conceptual Space refers to the categories and concept systems predefined by humans, providing a framework for AI to understand the world. However, relying solely on static conceptual frameworks, AI finds it difficult to cope with open and dynamic semantic scenarios, prone to "semantic misalignment" or even "hallucinations." For example, if intelligent agents with different language backgrounds have inconsistent internal representations of the same concept, misunderstandings and even decision deviations will occur. Professor Yucong Duan points out that the key lies in converting the symbolic Concept Space into the DIKWP Semantic Space to ensure consistent understanding of semantics by different subjects. In layman's terms, it is to transform human subjective conceptual models into semantic representations that machines can process, letting machines "read" the actual meaning behind human concepts.

In this transformation process, the DIKWP model plays a bridging role—it disassembles the concept hierarchy into five types of semantic elements: data, information, knowledge, wisdom, and Purpose, and provides a vertically connected coordinate system to map symbolic representations into a unified five-dimensional semantic vector. Through DIKWP semantic modeling, AI can understand and process concepts in a Unified Five-Dimensional Semantic Space, thereby breaking through the shackles of a single concept space and achieving precise alignment at the semantic level. In other words, in the semantic space, AI no longer just looks at the "name of the concept," but understands the multi-level semantic content associated with the concept and the Purpose behind it. In particular, the introduction of the Purpose Layer allows AI to refer to human goals and value considerations when making decisions, avoiding actions contrary to human original intentions. This is exactly the effect semantic sovereignty aims to achieve: making AI's internal semantic processing process Explainable and Controllable to humans, ensuring the realization of national will in the digital space. For example, when docking with an intelligent system in an ASEAN country, we need to ensure that both parties have a consistent semantic understanding and boundary for concepts such as "safety" and "integrity." This requires embedding the policy Purposes and cultural connotations of China and ASEAN on these concepts into the AI model through DIKWP modeling, achieving resonance at the semantic space level. Only in this way can cross-cultural AI cooperation avoid decision deviations caused by semantic misunderstandings and truly establish mutual trust.

(IV) Trustworthy AI and Embodied Intelligence: Future Directions

Trustworthiness and explainability are important features of Sovereign AI and intrinsic requirements for realizing semantic sovereignty. In addition to DIKWP white-box evaluation, we also need to comprehensively use various cutting-edge technologies to improve the trustworthiness and reliability of AI systems. One important trend is the development of Embodied AI. Embodied AI refers to endowing AI with a "body," allowing intelligent agents to actively learn and adapt through interaction with the physical world, thereby achieving a higher level of cognition. The "Embodied Intelligence Development Report" released at the 2024 World Robot Conference pointed out that embodied intelligence is a new paradigm of AI interdisciplinary integration, achieving intelligent growth through the interaction of physical entities with the environment. From the improvement of robot perception and action capabilities to the digital twin technology combining virtual and real, embodied intelligence will significantly enhance AI's ability to understand environmental context and human Purpose. This coincides with the goal of semantic sovereignty pursuing AI's deep grasp of human context and purpose.

Currently, domestic and foreign research institutions and enterprises are making embodied intelligence a new hot spot for AI development. Embodied intelligence technology is expected to be widely used in manufacturing, medical rehabilitation, education and training, and other fields. Its core lies in making AI "unity of knowledge and action"—both understanding semantics and being able to put them into action. For example, a service robot can only take appropriate action to serve if it understands the true Purpose behind the owner's language; autonomous vehicles need to constantly calibrate their semantic understanding of instructions and scenes through environmental interaction to ensure safety. It is foreseeable that the popularization of embodied intelligence will lift AI semantic understanding to a new level: AI will not only understand humans through data reasoning but also deepen its grasp of semantics through actual behavioral feedback. This closed loop of semantics and action will further consolidate the explainability and reliability of AI.

Professor Yucong Duan's team has begun to explore the research path of combining the DIKWP model with embodied intelligence. They proposed a framework for the neural mechanism of embodied intelligence integrating the DIKWP model, hoping to reveal how sensory information is transformed into knowledge representation, wise decision-making, and the role of Purpose therein, laying the foundation for higher-level machine cognition. This research is expected to reveal the mechanism of artificial intelligence from the sensorimotor layer to the semantic cognitive layer, providing theoretical support for future Brain-inspired Computing and artificial consciousness research. We believe that Guangxi should also layout the field of embodied intelligence in advance when promoting AI cooperation facing ASEAN: On one hand, rely on platforms like Guangxi Artificial Intelligence University to introduce and cultivate talents in interdisciplinary fields such as robotics and IoT, consolidating the foundation of embodied intelligence research; on the other hand, introduce embodied AI solutions in cooperation projects with ASEAN such as smart manufacturing and smart agriculture, driving technology iteration with practical applications. For example, deploying intelligent agricultural machinery in ASEAN agricultural parks, allowing AI to constantly optimize its understanding of crop growth semantics through interaction with the real environment, improving automated decision-making levels. By seizing the initiative in embodied intelligence, we can both enhance the originality of China's AI technology and provide AI products that better meet local needs for ASEAN, further consolidating cooperative relations.

In summary, we clarify the core concept of promoting Guangxi-ASEAN AI cooperation: Taking Semantic Sovereignty as the guiding principle, Sovereign AI as the goal, DIKWP model as the foundation, and Trustworthy Intelligence as the essential requirement.

Implementation Paths and Key Measures

Achieving trustworthy AI cooperation facing ASEAN requires not only innovation and integration at the technical level but also guarantee and coordination at the institutional level. As a pilot field, Guangxi should fully leverage its advantages, coordinate domestic and international resources, and explore replicable and scalable cross-border AI cooperation models. We suggest focusing on the following aspects:

(I) Building a Technical Integration Platform to Promote Localization of Achievements

Technical integration is the primary link in the "R&D in Beijing-Shanghai-Guangzhou + Integration in Guangxi + Application in ASEAN" model. Facing ASEAN's diverse geographical environments, languages, cultures, and infrastructures, it is necessary to screen, verify, and tune the AI foundation models and core algorithms received from first-tier cities to solve the problem of technology "acclimatization." To this end, Guangxi should rely on platforms like the China-ASEAN Artificial Intelligence Application Cooperation Center (Nanning A Center) to accelerate the construction of regional AI open-source communities and technical collaboration networks, attracting domestic and foreign developers to participate, sharing and reusing advanced models under a Controllable Framework, and innovating together. Specific measures include:

Introducing Domestic Top Technologies and Establishing Joint Laboratories. Encourage top AI enterprises from Beijing, Shanghai, and Shenzhen to set up branches or joint laboratories in Guangxi to pilot and optimize cutting-edge technologies locally. Typical examples include Beijing's Yunzhisheng Company, which has settled in the Nanning A Center, acting as a local "converter" to promote the landing of technologies from Beijing-Shanghai-Guangzhou in Guangxi and transforming them into applications needed by ASEAN. Another example is local enterprise Guangxi Xinchen Semiconductor, focusing on manufacturing production line landing, with its smart factory achieving local mass production of products, developing Internet of Vehicles and IoT intelligent hardware to serve Guangxi and explore the ASEAN market. These practices are effective models for re-innovating advanced technologies "according to local conditions."

Building China-ASEAN Trusted Data Zone and Public Service Cloud. Establish a Cross-border Data Sharing and AI Service Platform facing ASEAN in Guangxi, gathering multi-source heterogeneous data and model resources, providing one-stop solutions. Relying on this platform, mature AI capabilities such as Multilingual Machine Translation, Intelligent Customer Service, and Industrial Visual Inspection can be deployed for ASEAN users to call on demand. Through the "bag-in check-in" cloud service model, lower the threshold for ASEAN countries to adopt AI technologies. Specifically, build well Guangxi's characteristic advantage areas such as "Minority Language Machine Translation." A small smart translation card can achieve silky smooth conversion of multiple ASEAN languages, eliminating communication barriers. Currently, Guangxi enterprise Maiyue Technology has developed portable translation terminals, with R&D centers in Beijing-Shanghai-Guangzhou and Nanning, and is accelerating expansion into the ASEAN market. Support for the development of more such enterprises should be given to meet ASEAN's language service needs in business, tourism, education, and other scenarios.

Building Cross-border Computing Power and Network Infrastructure. Upgrade the construction of international communication gateways in Nanning and other places, improve network bandwidth and quality facing ASEAN, and guarantee real-time response of cross-border AI services. Explore establishing Regional Computing Power Centers in China-Vietnam border areas, Fangchenggang, etc., to provide cloud AI computing resources for ASEAN customers. Utilize Guangxi's rich clean energy advantages to build green data centers serving regional AI applications. At the same time, strive to include Guangxi in the coverage of the national "East Data West Computation" project, enhance computing power scheduling capabilities facing ASEAN, and provide hardware support for subsequent large model training and deployment.

Through the above measures, Guangxi will become an "integration transfer station" for domestic AI technological achievements, efficiently transforming source R&D potential into application momentum serving national strategy and facing the ASEAN market.

(II) Deepening Application Demonstration to Lead Scenario Implementation and Demand Traction

Application is the destination of AI value and the key grip for opening the ASEAN market. Guangxi should fully combine the development pain points and actual needs of ASEAN countries, select several key areas to carry out AI Application Demonstration Cooperation, and promote ASEAN market development from points to areas:

Smart Agriculture: Focus on the demand for digital upgrading of ASEAN traditional agriculture, using intelligent planting, drone field patrol, and agricultural product quality inspection as entry points. For example, in Malaysia, addressing the short harvesting window and extensive manual labor difficulties in tens of thousands of hectares of oil palm plantations, introduce the smart palm plantation solution of Guangxi enterprise Runjian Corp: using AI vision algorithms to identify fruit maturity, deploying drones equipped with hyperspectral cameras to mark mature fruits, cooperating with intelligent connected transport vehicles to navigate and harvest according to the map, not only reducing missed and early picking but also increasing yield per unit area by more than 10%. The president of the Malaysia Robotics Association evaluated this solution as a profound agricultural technology revolution. Suggest further promoting similar solutions in rubber, rice, and other planting industries in Thailand and Indonesia to improve quality and efficiency for ASEAN agriculture.

Smart Cities and Public Services: Combine management needs in ASEAN's urbanization process to create AI application templates in transportation, energy, medical care, and other fields. Currently, the China-Laos Artificial Intelligence Application Cooperation Center built by Guangxi Investment Group in Vientiane, Laos, has completed a display and interaction area of about 500 square meters, covering multiple modules from computing power infrastructure to Lao language large model R&D and enterprise credit investigation systems. Support should be given to continue deepening the construction of the China-Laos Cooperation Center and summarizing experiences to promote to Cambodia, Thailand, etc., forming a smart city cooperation network along the "Belt and Road." At the same time, encourage Guangxi technology enterprises to participate in smart port and smart medical projects in ASEAN countries. For example, using computer vision and IoT technology to transform the container dispatch system of Singapore ports, or providing intelligent diagnosis and health management platforms for Thai hospitals. These applications can meet local realistic needs and demonstrate the strength and responsible image of China's AI technology.

Cross-border Trade and Logistics: Relying on the deepening of the China-ASEAN Free Trade Area, create AI-driven cross-border trade facilitation services. Pilot Intelligent Customs Clearance Systems at ports like Pingxiang and Dongxing, using machine learning algorithms for cargo risk prediction and customs clearance process optimization, improving clearance efficiency and ensuring safety. Establish a Digital Supply Chain Platform facing ASEAN, integrating logistics information, optimizing transport routes and warehouse management through AI, and reducing cross-border logistics costs. Especially under the RCEP framework, cross-border e-commerce is booming. Guangxi can take the lead in developing Multilingual Cross-border E-commerce Intelligent Customer Service and Transaction Risk Monitoring systems, providing convenient services for SMEs in the region and enhancing discourse power in ASEAN trade.

Through successful application demonstrations in specific scenarios, we can verify the adaptability of China's AI technology in the ASEAN environment, establish a trustworthy brand, and drive further upgrading of ASEAN's demand for AI, forming a virtuous circle. Guangxi should leverage its advantage of being close to the frontline and market, keenly capture ASEAN cooperation opportunities, and use demonstration projects as a grip to leverage larger-scale cooperation.

(III) Strengthening Rule Alignment and Establishing Cross-border Semantic Governance Mechanisms

As technology applications cross borders, rules and standards need to be aligned synchronously. For cross-border AI cooperation to go far and steady, there must be a corresponding institutional cornerstone. We must regulate during development and develop during regulation, Realizing "Agile Governance", responding quickly to technological evolution through dynamic adjustments. Specific suggestions:

Co-building China-ASEAN AI Governance Collaboration Platform. Propose that China's cyberspace administration departments take the lead, unite with competent authorities and university think tanks in ASEAN countries, to establish the "China-ASEAN AI Governance and Standards Joint Research Center." This center can conduct collaborative research on issues such as Cross-border Data Flow, Algorithmic Ethics, Security Risk Control, publish regional AI governance reports annually, and propose rule initiatives reflecting common regional interests. Through this platform, regional AI governance guidelines can be gradually formed, such as data sharing and privacy protection guidelines, multilingual content review standards, algorithm transparency evaluation indicators, etc., providing an "Asian Example" for future international rules.

Exploring "Sandbox Supervision" and Risk Emergency Response. In Guangxi Free Trade Zone or Nanning, pilot the AI Application Sandbox Supervision mechanism: allow emerging cross-border AI services to be tested and deployed within a limited scope, while regulators and third-party assessment teams continuously monitor and evaluate them. Based on sandbox feedback, adjust rule details in time to ensure innovation is not stifled while risks are prevented. At the same time, plan to establish a Cross-border AI Dispute Consultation and Emergency Disposal Platform. In case AI services cause cross-border disputes (such as misleading content, data leaks, etc.), multilateral consultation and emergency response can be quickly initiated to reduce the impact of the event. This platform can dock with mechanisms like the China-ASEAN Cybersecurity Emergency Hotline to jointly improve regional risk joint prevention and control capabilities.

Promoting Standard Mutual Recognition and Policy Docking. Actively dock with AI-related policies already formulated or proposed by ASEAN, and discuss the possibility of Standard Mutual Recognition with ASEAN partners. Documents such as the "Standard for Artificial Intelligence Semantic Transparency" drafted under China's lead (detailed in the policy recommendation section later) can invite ASEAN countries to participate in formulation, striving to make them bilateral or even multilateral standards. Similarly, China's advocacy for "Data Security" and "Digital Ethics" should also be integrated into the ASEAN digital governance framework. Consider writing the principle of "Multilingual Semantic Interoperability" into the upcoming "China-ASEAN Digital Economy Cooperation Action Plan" or RCEP extension agreement annexes, clarifying commitments of all parties in AI semantic consistency and cultural respect in the form of policy. This will improve the uniformity and predictability of regional rules, giving enterprises rules to follow when providing cross-border AI services.

By perfecting the above mechanisms, Guangxi's AI cooperation with ASEAN will be built on a more solid foundation of rules. An Agile, Collaborative, and Mutual Trust cross-border AI governance framework not only guarantees cooperation security but also creates conditions for China to win initiative in international rule-making.

(IV) Consolidating Multilingual Resources and Talent Support

Multilingual capability is Guangxi's trump card in serving ASEAN AI cooperation. We must seize this advantage and make great efforts to build Multilingual Data Resource Repositories and Composite Talent Teams:

Building Multilingual Corpus and Knowledge Base. Suggest national support to establish the China-ASEAN Multilingual Corpus Center in Guangxi, gathering massive high-quality text, voice, and video corpora including Chinese and major ASEAN languages. The center should cooperate with domestic research institutions and enterprises, using Semantic Labeling and Alignment technologies to clean, tag, and structure the corpus, forming a standardized corpus usable for training multilingual large models. At the same time, construct a Multilingual Knowledge Graph, gathering general knowledge and local knowledge from various countries, achieving associative mapping between knowledge in different languages. This can not only improve AI's mastery of knowledge in different cultural backgrounds but also help discover cross-cultural semantic differences, providing a basis for model correction. Through the construction of multilingual data resources, we can ensure AI development conforms to the value requirements of China and partner countries, and master more discourse power in global AI governance. Specifically, given that Chinese accounts for a negligible proportion of international AI training data, we should accelerate the construction of Large-scale Chinese and Minority Language Corpora, strengthen data sharing and opening, to change this unfavorable situation. The Third Plenary Session of the 20th Central Committee emphasized enhancing cultural confidence and inheriting excellent traditional Chinese culture. We can fully incorporate Chinese cultural materials into the corpus and learn from OpenAI's experience to create Chinese GPT series models, enhancing the dissemination power of Chinese culture in the global digital space. Data security and privacy protection must also be noted in this process to ensure data resources are usable but not abused.

Conducting China-ASEAN AI Talent Joint Training. Utilizing platforms like Guangxi Artificial Intelligence University, implement the "China-ASEAN AI Talent Training Plan." Models such as joint university operations and joint enterprise training can be adopted to select a certain number of ASEAN students to come to Guangxi for advanced studies in AI-related majors annually, while selecting outstanding Guangxi students to exchange and study in ASEAN, cultivating composite talents who master AI technology and are familiar with ASEAN languages and cultures. Education and technology departments can jointly support the landing of projects like the Embodied Intelligence Industry-Education Integration Innovation Center in Guangxi, cultivating skilled talents in robotics and automation urgently needed by ASEAN countries. In addition, establish China-ASEAN Artificial Intelligence College Branches or training bases in cities like Nanning, conducting short-term training classes for government officials and industrial technical personnel. In November this year, the China-ASEAN Artificial Intelligence Innovation Cooperation Training Class hosted by the Ministry of Commerce was successfully held in Nanning, Guangxi. Officials and scholars from 5 countries including Cambodia, Indonesia, and Malaysia conducted in-depth exchanges on AI large model applications and digital economy cooperation. Such training should be normalized and institutionalized, forming annual regular talent training projects to improve regional overall talent literacy and technology application capabilities. Through localized training, Guangxi is expected to become an AI Talent Highland serving ASEAN, providing a steady stream of human resources for the long-term development of cooperation.

(V) Shaping Cooperative Brands and Expanding International Influence

While promoting specific cooperation, attention should also be paid to shaping the brand image of China-ASEAN AI cooperation and enhancing international discourse power:

Creating High-end Exchange and Cooperation Platforms. Relying on major events like the Nanning Garden Expo and the China-ASEAN Expo, add "AI Cooperation" special forums or exhibition zones, inviting ASEAN science and technology ministers, renowned experts, and entrepreneurs to exchange annually, displaying the latest cooperation results. Through continuous public diplomacy, make "Look to Guangxi for AI Cooperation" a well-known brand. In 2025, Guangxi successfully hosted the first China-ASEAN AI Ministerial Roundtable. Suggest solidifying this mechanism, holding it alternately in China and ASEAN countries annually, forming a long-term mechanism for high-level dialogue. At the same time, strive to list Chinese initiatives such as "Semantic Sovereignty" as one of the meeting topics, better conveying China's concepts to ASEAN national leaderships to win understanding and support.

Conducting Joint Research and Competitions. Support Guangxi universities and research institutions to take the lead in co-building "Belt and Road AI Joint Laboratories" with ASEAN universities, jointly applying for research topics, publishing papers and monographs, and improving academic influence. Hold activities like the China-ASEAN AI Competition (similar to the "A Super" League), inviting outstanding ASEAN teams to participate, promoting research through competitions, gathering talents through competitions, and discovering cooperation potential. For ASEAN projects and teams performing prominently in competitions, provide policy support for incubation and landing in Guangxi, making the competition a link for technology exchange and project introduction.

Strengthening Publicity and Public Opinion Guidance. Fully utilize central media like People's Daily, Xinhua News Agency, and mainstream ASEAN media to publicize and report successful cases, character stories, and social benefits of China-ASEAN AI cooperation, creating a good public opinion atmosphere. Especially for the concept of semantic sovereignty and its practical results, voice out through professional media and international conferences, clarify misunderstandings, and strive for international recognition. Organize the translation and publication of the Multilingual Version of the "White Paper on Semantic Sovereignty", systematically introducing China's policy propositions and progress in semantic governance, distributed to decision-makers and institutions of various countries for reference. Through planned publicity, gradually establish China's international image as a Responsible AI Major Power and a Pioneer in Semantic Governance.

In summary, Guangxi's promotion of AI cooperation facing ASEAN should, under central coordination, build platforms and lead demonstrations technically, perfect rules and cultivate talents institutionally, and establish brands and lead discourse externally. Through the combined effect of these implementation paths, we are confident in building Guangxi into a Pioneer Practice Zone for the concepts of Sovereign AI and Semantic Sovereignty, providing beneficial experiences for the whole country and even the world.

Policy Recommendations

To ensure the smooth realization of the above goals, strong support is needed at the national strategic and policy levels. We propose the following policy recommendations:

Incorporate "Semantic Sovereignty" into the Top-level Design of National AI Strategy. Suggest that the Cyberspace Administration of China, Ministry of Science and Technology, Ministry of Industry and Information Technology, National Data Bureau, etc., take the lead in adding content related to "Semantic Sovereignty" when revising programmatic documents such as the "New Generation Artificial Intelligence Development Plan" and "Digital China Construction Plan," clarifying its strategic status and development path. Specific measures include: setting up a Semantic Governance Special Working Group under the National AI Governance Committee to coordinate various departments to promote the implementation of semantic sovereignty; including semantic sovereignty in the assessment indicators of Digital China and Cyber Power strategies, requiring governments at all levels to add Semantic Security Assessment links in smart city and digital government construction to implement semantic sovereignty requirements; listing semantic sovereignty as a special topic at the annual National Science and Technology Work Conference and Digital Economy Promotion Conference to exchange progress and issues, maintaining high-level attention. Through the above top-level design arrangements, elevate semantic sovereignty to the status of a New Element of Digital Sovereignty as important as data sovereignty and cyber sovereignty, providing directional guidance and organizational guarantees for subsequent work.

Establish National Major Projects for Semantic Sovereignty to Increase Key Technology Breakthroughs. To consolidate the technical foundation of semantic sovereignty, suggest adding a "Key Technologies for Semantic Sovereignty" key R&D special project under the national science and technology plan. Organized and implemented by the Ministry of Science and Technology in conjunction with the Natural Science Foundation, focusing on the following tasks:

Advanced Semantic Modeling Theory: Support new theoretical research on DIKWP semantic mathematical models and semantic representation learning, exploring mathematical principles and cognitive mechanisms of machine semantic understanding, providing long-term theoretical supply for semantic sovereignty.

Artificial Consciousness White-box Evaluation Tools: Perfect the AI white-box evaluation toolset, expanding the coverage of evaluation question banks. Design evaluation standards and indicator systems from multiple dimensions such as perception, language, ethics, and Purpose, forming an International Benchmark for Multilingual White-box Evaluation, with China initiating the formulation of evaluation standards. Through the R&D and promotion of evaluation tools, master the initiative in AI capability assessment.

Semantic Blockchain and Trusted Data Circulation: Tackle performance and security issues of semantic blockchain, developing efficient semantic evidence storage algorithms. Explore using blockchain for the confirmation and circulation of data elements and knowledge: every data record on the chain carries DIKWP five-layer semantic metadata, realizing a Traceable, Confirmable, Machine-understandable decentralized semantic ledger. Pilot the construction of a DIKWP semantic data exchange platform in Hainan Free Trade Port or Guangxi where conditions are mature. Through technical special projects, strive to break through semantic blockchain technology bottlenecks, embedding semantic sovereignty requirements for cross-border data flow, ensuring the authenticity and consistency of cross-border data exchange.

DIKWPaaS Platform and Industry Demonstration: Develop Semantic Sovereignty Platform as a Service (DIKWPaaS), modularizing the five-layer functions of the DIKWP model into cloud services for various industries to call on demand. Support industry-university-research collaboration to create benchmark applications in smart government, financial risk control, medical knowledge engineering, etc., forming replicable and scalable industry solutions. For example, R&D of a Chinese-led Global Knowledge Graph to enhance China's control over international semantic resources; develop an Autonomous and Controllable Semantic Operating System to serve information systems in government and key industries, ensuring semantic consistency from the underlying operating environment to upper-layer applications.

The special project should encourage attracting top domestic teams for joint tackling through mechanisms like "Revealing the榜 and Taking Command," supporting leading units like Professor Yucong Duan's team at Hainan University to lead the application, uniting head enterprises like Huawei and Baidu, and universities like Chinese Academy of Sciences and Tsinghua University to co-build a "Semantic Sovereignty Technology Innovation Alliance." In terms of funding, suggest investing special funds annually under the "14th Five-Year" National Key R&D Plan for experimental facilities, data resources, and talent cultivation. After 3-5 years of tackling, strive to form a batch of internationally influential semantic sovereignty scientific research results and patent layouts (Professor Yucong Duan's team has currently been granted 114 related invention patents, possessing a good foundation), enhancing China's independent innovation capability and discourse power in AI core technologies. These results are not only the embodiment of China's AI soft power but will also feed back into national security and industrial upgrading.

Promote the Construction of Semantic Sovereignty Standard Systems to Strive for Dual International and Domestic Discourse Power. Standards are the commanding heights of emerging technology competition. To consolidate semantic sovereignty achievements, it is necessary to normalize and standardize key technologies and practices. On one hand, domestically, the Standardization Administration of China should take the lead, in conjunction with the MIIT, CAC, etc., to formulate the National Semantic Sovereignty Standard System Plan. Prioritize the development of the following standards:

"Technical Specifications for DIKWP Semantic Model": Prescribe technical requirements and interface forms for the DIKWP five-layer model, guiding various AI systems to adopt a unified semantic framework.

"Technical Requirements for Artificial Intelligence White-box Evaluation": Clarify methods and indicators for AI semantic assessment as mandatory evaluation standards before AI systems go online, ensuring their semantic decision-making processes are transparent and compliant.

"Semantic Blockchain Data Format and Interface Specifications": Unify semantic blockchain data structures, metadata descriptions, and calling interfaces to achieve cross-platform interconnection.

"Multilingual Semantic Interoperability Specifications": Prescribe technical requirements for semantic conversion and alignment between multilingual AI systems, guaranteeing semantic consistency and unambiguity when different language AI services collaborate.

"AI Semantic Security Assessment Guidelines": Formulate method guidelines for identifying and preventing semantic bias and semantic attacks, providing reference for regulators.

The above standards can first be released as recommended national standards, and rise to mandatory standards or industry access requirements when conditions are mature. Especially in fields involving ideological security and public services (such as News AI, Education AI, Judicial AI), semantic sovereignty standards should be hard constraints, regularly testing and evaluating the semantic transparency and value orientation of AI systems, ensuring that AI systems operating in China meet basic semantic security indicators. On the other hand, internationally, China should proactively Export Chinese Solutions to stages like ISO, IEC, ITU. Utilizing China's influence in organizations like ISO/IEC JTC1 (Information Technology) and ITU-T, promote the establishment of an "AI Semantic and Knowledge Governance" working group, with Chinese experts leading the formulation of international standard proposals. For example, the aforementioned "Semantic Transparency for AI Systems" can be a Chinese proposal, incorporating DIKWP model ideas and China's industrial practice data. At the same time, strive to write the principle of "Multilingual Semantic Interoperability" advocated jointly by China and ASEAN partners into UN digital governance initiatives or documents, increasing international recognition. In the standard formulation process, widely unite countries along the "Belt and Road" to participate, forming a "China-led, Multi-country Supported" standard alliance to avoid Western countries monopolizing standard discourse power. Through dual-track promotion domestically and internationally, strive to have a basically sound semantic sovereignty standard system comprehensively implemented domestically around 2030, with several standards becoming ISO/IEC international standards adopted globally. The landing of standards will further consolidate China's leading position in semantic sovereignty technology and governance.

Improve Laws and Regulations to Provide Institutional Guarantees for Semantic Governance. The standardized development of AI cannot do without legal escort. Suggest the National People's Congress and the CPPCC play a role in legislative suggestions, accelerating the layout of semantic governance content in the AI Legal Framework:

Revise and Perfect Existing Laws: Launch research on revising laws such as the "Cybersecurity Law," "Data Security Law," and "Personal Information Protection Law," adding AI semantic security and cultural security clauses. For example, explicitly stipulate in the "Data Security Law" that "The State maintains sovereign rights and interests at the level of artificial intelligence semantics, prohibiting any organization from using algorithms to distort or tamper with Chinese cultural semantics"; add requirements in the "Cybersecurity Law" that "Operators of critical information infrastructure shall ensure that the semantic decision-making process of their artificial intelligence services is explainable and auditable." By modifying higher-level laws, fix the principle of semantic sovereignty in legal form, making it a bottom line that the AI industry and platform enterprises must observe.

Formulate Special Regulations or Guidance Opinions: Research and issue "Artificial Intelligence Semantic Governance Regulations" or State Council guidance opinions, drafted and released under the lead of the CAC. This regulation can clarify the semantic compliance obligations of AI product and service providers (such as regularly conducting white-box evaluations, establishing semantic bias correction mechanisms, etc.), clarify the division of responsibilities of various industry regulators (e.g., NRTA regulating semantic security of media AI content, Ministry of Education regulating value orientation of education AI, etc.), and violation penalties (e.g., accountability measures for serious consequences caused by AI system semantic misleading). Fill current legal gaps in the form of administrative regulations, making supervision law-based and rule-based. Meanwhile, explore including AI semantic bias, discrimination, etc., into the scope of judicial review in judicial cases, providing relief channels for victims, and constantly perfecting the guarantee of semantic security at the judicial level.

Pre-research Artificial Intelligence (Including Semantic Governance) Basic Legislation: List the "Artificial Intelligence Law" in the medium and long-term legislative plan. Given that the EU launched the "AI Act" first in 2024, and China also implemented the "Interim Measures for the Management of Generative Artificial Intelligence Services" in August 2023 and included the "Artificial Intelligence Law" in the 2024 legislative deliberation plan. When formulating comprehensive AI laws, fully integrate semantic sovereignty characteristics, coordinate domestic and international situations, balance innovation and security, preventing high risks while supporting legitimate innovation. This law should reflect Chinese wisdom, drawing on international experience while highlighting our cultural values, covering AI R&D, data governance, algorithmic ethics, semantic security, talent cultivation, etc., providing long-term legal guarantees for China's AI development. Through the perfection of basic legislation, ensure semantic governance and Sovereign AI construction run healthily on the track of the rule of law.

Support Guangxi in Piloting and Creating a Model for Semantic Sovereignty Practice. The central government can grant Guangxi special policies for Reform Experiments in the fields of semantic sovereignty and Sovereign AI, encouraging Guangxi to take the lead in exploring replicable experiences:

Regional Pilot Policies: Suggest including Guangxi in the national new generation AI innovation development pilot zone, giving inclination in funds, projects, and talents, supporting Guangxi in exploring cross-border data flow security, AI ethics review mechanisms, etc. For example, authorize Guangxi to conduct Cross-border Data Sandbox Pilots in Nanning, Qinzhou, etc., allowing some semantic blockchain and data cross-border circulation projects to be practiced first within the regulatory sandbox, and promoting nationwide after summary and evaluation.

National Platform Layout: Support the establishment of national-level scientific research platforms in Guangxi, such as the National Semantic Intelligence Engineering Technology Research Center (Guangxi), "Belt and Road Artificial Intelligence Innovation Center," etc., gathering domestic top intellectual resources to serve Guangxi practice in the form of projects and topics. Support Guangxi Artificial Intelligence University in establishing high-level academic journals or conferences, creating an internationally renowned academic exchange base, enhancing Guangxi's discourse power in the AI academic community.

Assessment and Incentive Mechanism: Include Guangxi's exploration results in semantic sovereignty construction into the assessment content of Digital China and Cyber Power, commending and promoting those with significant results. For China-ASEAN AI Cooperation Standards and governance paradigms exported by Guangxi, encourage relevant national departments to adopt and learn from them in related fields nationwide, enhancing the value of Guangxi's reform pilots. Give certain inclination in rating and promotion to Guangxi cadres and talents who act actively and innovate boldly, stimulating local motivation for entrepreneurship.

Through the above policy support, Guangxi's exploration will be more confident, motivated, and directional, accumulating valuable experience for the whole country while accelerating its own development, truly undertaking the mission of "AI Cooperation Highland Facing ASEAN" endowed by the state.

Strengthen International Cooperation to Build a Digital Model for a Community with a Shared Future for Mankind. Semantic sovereignty is not only a concept advocated by China but also fits the common demands of the vast number of developing countries to maintain digital sovereignty. Guangxi should be guided to deepen collaboration with ASEAN and countries along the "Belt and Road" in AI governance and standards based on existing cooperation, jointly promoting Global Digital Governance Transformation:

Expand Multilateral Cooperation: Propose the "Global Semantic Equality" initiative under the G77 and China-ASEAN frameworks, advocating countries to jointly oppose algorithmic colonialism and promote the equal coexistence of multilingual cultures in the AI era. Support discussions on "Artificial Intelligence and Cultural Diversity" in organizations like UNESCO and ITU, actively share China's experience in promoting semantic sovereignty, and strive for response from more countries. Through multilateral occasions, China can join hands with ASEAN countries to voice out, promoting the inclusion of semantic sovereignty as part of the UN digital agenda, providing China-ASEAN solutions for formulating global rules.

Deepen Bilateral Cooperation: In addition to ASEAN as a whole, select key countries to sign AI Semantic Governance Bilateral Cooperation Memorandums. For example, establish joint working groups with representative countries in population and language like Indonesia and Malaysia, strengthening information exchange and capacity building on specific issues like data sharing, algorithm review, and content regulation. Encourage Chinese enterprises to set up AI Compliance Innovation Centers in these countries, helping cultivate local algorithm governance talents, and exporting China's mature semantic governance tools (such as the DIKWP white-box evaluation platform) for their use. This can both enhance mutual trust and help Chinese technical standards better integrate into their systems.

Co-build "Belt and Road" Digital Trust System: Leveraging the "Belt and Road" initiative, advocate building an open and inclusive Digital Trust System. The core is supported by White-box Evaluation, Blockchain Evidence Storage, and Multilateral Audit, ensuring cross-border data and AI services flow under a framework of traceable sources, trusted content, and controllable processes. Launch the "DIKWP Semantic Compliance Label" system first along the "Belt and Road"—similar to ISO quality certification, assessing the semantic consistency of smart city AI platforms through quantitative indicators, granting certification labels to compliant systems as a pass to enter various markets. If this initiative is recognized by countries along the route, it will greatly enhance China's institutional discourse power in global AI governance and provide practical support for building a community with a shared future in cyberspace.

The above policy recommendations are based on current challenges and future trends, focusing on both short-term feasible measures and long-term institutional frameworks, aiming to comprehensively guarantee the smooth promotion and sustainable development of Guangxi's AI cooperation facing ASEAN.

Expected Outcomes

Under the strong leadership of the central government and the promotion of the above measures, we expect to achieve the following positive results:

First, cross-border AI exchanges will be more trustworthy and controllable. Through the implementation of the semantic sovereignty framework, the problems of semantic transparency and value consistency between Chinese and foreign AI systems will be solved. Regulators and the public in various countries can clearly understand the basis and Purpose behind AI decisions, eliminating doubts about "algorithm black boxes." Cultural misreadings and biases in AI output content will be significantly reduced, achieving fair and accurate communication between different languages and diverse cultures. For example, multilingual communication channels capable of real-time translation and value judgment will be built between China and ASEAN countries, ensuring "no distortion" and "no misreading" in both business dialogues and policy exchanges. This will significantly improve the efficiency and trust of cross-border exchanges, creating a more harmonious public opinion environment for co-building the "Belt and Road."

Second, cross-border flow of data and knowledge will be safe and efficient. With technologies like semantic blockchain as escort, future cross-border data circulation will balance security and efficiency. The entire process of data from collection, transmission to use will carry meta-information such as context, knowledge, and Purpose recorded on the chain. This allows regulatory authorities to trace data sources and their evolution into decision-making links in real-time. Countries can safely share data for AI model training and federated learning based on commonly recognized semantic labeling standards, thereby breaking the "data silo" dilemma. For example, in the medical field, hospitals in China and Singapore can share patient imaging data to train AI diagnostic models while ensuring all data usage is limited to agreed medical Purposes (labeled via P layer) and is not distorted. This not only improves data utilization efficiency but also guarantees privacy and sovereignty. In the long run, a Trusted Data Space covering China-ASEAN will gradually form, injecting strong impetus into digital trade and cross-border finance.

Third, the AI industry will achieve mutually beneficial development. Through Guangxi's bridge role, China's AI industry will open up the broad ASEAN market, and ASEAN will also share the dividends of China's AI development. It is estimated that by 2027, the output value of Guangxi's AI-related industries facing ASEAN is expected to exceed the planned target of 100 billion yuan, forming several characteristic industrial clusters with regional influence. Technology Spillover Effects will cause a batch of AI application enterprises and start-up teams to emerge locally in ASEAN countries, making the regional AI ecosystem more prosperous. Chinese enterprises gain valuable data resources and application feedback in the process of serving ASEAN, feeding back into their own technology iteration. A batch of AI products represented by translators, intelligent customer service, and agricultural robots will become star achievements of China-ASEAN cooperation. More importantly, by jointly participating in formulating standards and regulations, Chinese and foreign enterprises will compete and cooperate under a Unified and Transparent rule system, making the market order more benign and healthy. This lays the foundation for deepening Digital Economy Cooperation under the RCEP framework.

Fourth, national cultural soft power and international discourse power are significantly enhanced. The implementation of semantic sovereignty ensures that China's mainstream values and culture are promoted and disseminated in the AI era. Our language and culture will no longer be submerged and distorted by external algorithms but will go global in digital form. With the advancement of Chinese corpus and bilingual large model construction, Chinese cultural classics and development concepts can be output to global users through generative AI products, shaping a more positive image of China in the international public opinion field. ASEAN countries' understanding of Chinese culture will be deeper, consolidating the foundation for people-to-people bonds. On the other hand, China's discourse power on the international AI governance stage will be significantly improved—concepts we proposed translate into international standards, models we developed become tools adopted by countries, and rules we advocated are widely recognized and embedded in global governance documents. When more and more countries "look up to" China's experience in semantic governance, China will undoubtedly stand in the leading position of global digital governance. In future international competition, this Institutional Discourse Power will become a core competitiveness comparable to technological strength.

Fifth, the safe and controllable development environment is more consolidated. Through multiple means such as laws, standards, and evaluations, the risks of AI development will be effectively controlled. The deployment and operation of AI systems in China will always remain in a regulated and auditable state. Any attempts to use algorithms to distort China's policies or spread bad ideologies can be discovered and prevented in time (e.g., discovering model output deviations through regular semantic assessment and correcting them). AI services relied upon by China's critical information infrastructure will also be easier to overhaul and troubleshoot due to semantic transparency, enhancing security. At the same time, the people's trust in AI will increase, and social psychological expectations will be stable. While intelligent technology dividends are released in more industries, disputes and accidents caused by cognitive misleading will be significantly reduced. Overall, we will accelerate the landing of large-scale AI applications under the premise of ensuring safety, realizing a virtuous cycle of technology for good and industry for the better.

Sixth, a cooperative and win-win digital community with a shared future is taking shape faster. Through innovative cooperation with ASEAN in the AI field, we have explored a new path for developing countries to coordinate Digital Autonomy and Open Cooperation. This model proves that: based on respecting the digital sovereignty of each country, it is still entirely possible to realize the cross-border flow and sharing of technology and data, thereby achieving win-win cooperation. It shows the world a choice distinct from Western monopolies and closed barriers, demonstrating China's sincerity and responsibility in building a community with a shared future for mankind. It is believed that with the maturity of this model, not only ASEAN but more countries in Latin America, Africa, and the Middle East will join in the future, forming a larger-scale digital cooperation network. China will work with partner countries to jointly shape a Diverse Coexistence, Mutual Trust and Benefit new order of global digital governance. This will be a powerful response to unilateralism and digital hegemony, injecting new impetus into world peace and development.

In summary, by strengthening top-level design and policy support, and using Guangxi as a pilot to deepen AI cooperation facing ASEAN, we are confident in realizing the Organic Unity of technological progress, industrial development, and security governance, benefiting regional people while serving national strategy. Guangxi's practical experience will also provide a demonstration for the whole country and contribute wisdom to the world. We sincerely hope the central government can give attention and support, and ask leading comrades to indicate the direction of the next step of work.

Appendix: Summary of Attachment Content

(Attachment: Proposal on Promoting the Inclusion of "Semantic Sovereignty" into the National Artificial Intelligence Strategic System, November 2025)

I. Background and Problems: Current AI governance faces risks such as "semantic black box," "semantic bias," and "value misreading," highlighting challenges to cultural security and discourse power. Countries are exploring new concepts, and "Semantic Sovereignty" has emerged, aiming to maintain national digital context autonomy.

II. Theoretical Basis and Support: Semantic Sovereignty was proposed by Professor Yucong Duan, defined as the state's right and ability to ensure its own language, culture, and values are accurately and completely expressed in the digital environment. This concept expands the connotation of traditional data sovereignty, emphasizing control over the "meaning" carried by data. To realize semantic sovereignty, Professor Duan's team's original DIKWP Model is introduced as a technical cornerstone. DIKWP expands the classic DIKW hierarchy, adding a "Purpose" layer, constructing a complete cognitive chain of Data-Information-Knowledge-Wisdom-Purpose. Through this model, AI cognition can be ensured to integrate into human goal contexts, providing a foundation for semantic autonomy. Professor Duan's team developed the DIKWP White-box Evaluation System, used to assess AI capabilities at various cognitive levels to ensure Sovereign AI conforms to human Purposes. At the Data Factor Forum in July 2025, Professor Duan called for breaking through black-box evaluation and establishing white-box evaluation standards, having released corresponding methods, hoping to become international AI assessment standards.

The proposal points out that realizing semantic sovereignty requires AI to move from static Concept Space to dynamic Semantic Space. Traditional Concept Space is difficult to cope with open contexts, easily leading to semantic misalignment. The key is to convert symbolic concepts into DIKWP semantic representations, ensuring consistent semantic understanding by different intelligent agents. DIKWP provides five-dimensional semantic coordinates, decomposing concepts into Data, Information, Knowledge, Wisdom, and Purpose elements, allowing AI to process concepts in a unified semantic space, achieving semantic alignment. Especially the introduction of the Purpose layer makes AI decisions consider human purposes, avoiding deviation from original intentions. This makes the AI semantic process explainable and controllable, guaranteeing the realization of national will.

III. Application Scenarios and Pilots: The proposal suggests practicing the concept of semantic sovereignty first in fields like smart cities, digital government, and cultural dissemination. For example: constructing a Semantically Labeled Government Knowledge Base to improve the accuracy of government AI interpretation of policies; applying white-box evaluation in media AI to prevent value deviations; selecting several cities to carry out Semantic Blockchain Data Exchange pilots, recording data context and usage when public data is opened, ensuring semantics are not distorted during sharing. Experts have currently suggested testing the DIKWP semantic data platform in Hainan Free Trade Port. The proposal encourages testing semantic sovereignty technology and management mechanisms through pilot demonstrations to accumulate experience for nationwide promotion.

IV. International Competition and Discourse Power: Semantic sovereignty represents the new frontier of global digital governance and is a new grip for building China's international discourse power. The proposal analyzes the advantages of Europe and America in AI standards and value output, pointing out that China should actively shape rules to avoid being controlled by others. The team led by Professor Yucong Duan actively participates in ISO standards, successfully incorporating the DIKWP model into international smart city standards. To continue seizing the commanding heights of standards, the proposal advocates establishing an International AI Semantic Standard Alliance, led by China and uniting developing countries to jointly draft standards for AI semantic transparency and semantic security, proposing multilingual semantic interoperability initiatives at the UN and other occasions, building a "China-led, Multi-country Supported" pattern to avoid Western monopoly on standards. Through standard and rule output, enhance China's institutional influence in global AI governance.

V. Policy Recommendations: To integrate semantic sovereignty into China's AI strategic system, the proposal puts forward the following policy measures:

Top-level Design: Include semantic sovereignty in national AI strategic documents, establish a cross-departmental working group, include in Digital China assessment indicators, and discuss as a special topic at high-level meetings annually to enhance attention.

Major Projects: Establish national key technology special projects for semantic sovereignty, focusing on DIKWP semantic mathematics, white-box evaluation tools, semantic blockchain, DIKWPaaS, etc., supporting Professor Yucong Duan's team and others to lead, uniting industry-university-research to create core technologies and demonstration systems. Suggest annual special funding under the "14th Five-Year" Key R&D Plan to produce a batch of core patents and results.

Standard System: Formulate national semantic sovereignty standard system planning, prioritizing development of standards for DIKWP model, AI white-box evaluation, semantic blockchain, multilingual interoperability, semantic security assessment, etc. Standards should first be recommended national standards, then rise to mandatory upon maturity, especially as hard constraints in media, education, judicial fields, regularly tested. Simultaneously initiate international AI semantic standards in ISO/IEC, such as "AI Semantic Transparency," and write multilingual semantic interoperability into UN documents. Unite "Belt and Road" countries to form a standard alliance, striving for a perfect domestic standard system around 2030, with some becoming international standards.

Legal Guarantee: Promote amendments to the "Cybersecurity Law," "Data Security Law," etc., adding AI semantic security clauses, such as prohibiting algorithms from tampering with Chinese semantics, requiring explainability of critical institution AI decisions. Formulate "Artificial Intelligence Semantic Governance Regulations," clarifying AI service provider semantic compliance obligations (white-box evaluation, correction mechanisms, etc.), regulator responsibilities (NRTA managing media AI semantics, etc.), and violation penalties. Incubate the "Artificial Intelligence Law," incorporating semantic governance, rising to higher law. Establish semantic sovereignty principles through legal means, making AI development and application law-based and rule-based.

Organizational Guarantee: Strengthen overall coordination, clarify leading units and division of labor mechanisms. Suggest the CAC lead overall promotion, MOST responsible for technical breakthroughs, MIIT for standard industrialization, MOE for talent cultivation, propaganda departments for content supervision, and CMC Science and Technology Commission focusing on military AI semantic security, etc. Establish cross-departmental collaborative working mechanisms, regularly consult on progress, and solve problems in time. Encourage localities (such as Hainan, Guangxi) to carry out pioneering trials, with central government giving policy support and fault tolerance incentives to form promotable experiences.

The above proposal content provides detailed reference for our policy formulation. The main body of this report has fully absorbed reasonable suggestions therein and expanded them in combination with the reality of Guangxi's cooperation facing ASEAN. The above appendix summary is for leading comrades' reference.

References

Guangxi Builds AI Cooperation Highland Facing ASEAN | Reference News Special | Guangxi_Sina Finance_Sina.com, https://finance.sina.com.cn/roll/2025-10-23/doc-infuwnhi2338690.shtml

Guangxi Artificial Intelligence University 2025 Opening Ceremony and Academic Salon Held (Photo), https://www.guet.edu.cn/2025/0910/c9041a141605/page.htm

(PDF) Proposal on Promoting the Inclusion of "Semantic Sovereignty" into the National Artificial Intelligence Strategic System, https://www.researchgate.net/publication/393461651_guanyutuidongyuyizhuquannaruguojiarengongzhinengzhanetixidetian

Strengthening "Sovereign AI" Construction, Enhancing Competitiveness - Think Tank Forum - People's Weekly Network—Mainstream Public Opinion Convergence Communication Platform, https://www.peopleweekly.cn/html/2024/renminzhiku_1111/230288.html

(PDF) Constructing Sovereign AI Semantic System: Semantic Sovereignty and DIKWP Path Deepening, https://www.researchgate.net/publication/393464519_goujianzhuquanAIyuyitixiyuyizhuquanyuDIKWPlujingshenhua

Detailed Explanation of Semantic Sovereignty: A New Topic in Global Digital Governance - Zhihu Column, https://zhuanlan.zhihu.com/p/1920150990931341732

(PDF) DIKWP Model Included in ISO/IEC TR 25005-2:2025 Smart City International Standard, https://www.researchgate.net/publication/396929275_DIKWP_Model_Included_in_ISOIEC_TR_25005-22025_Smart_City_International_Standard

ScienceNet—Chapter 1: Global Standard Landscape and DIKWP's Strategic Positioning- Yucong Duan's Blog, http://https--blog--sciencenet--cn.proxy.xianning.gov.cn/blog-3429562-1508101.html

Embodied Intelligence Development Report, https://hulianhutongshequ.cn/upload/tank/report/2024/202408/1/a3049be804ae4b9eb690bd7e798da62e.pdf

CAICT Jointly Released "Embodied Intelligence Development Report (2024)" - Sina Finance, https://finance.sina.com.cn/tech/roll/2024-08-28/doc-incmczxu3786325.shtml

(PDF) Call for Cooperation: Research on Embodied Intelligence Neural Mechanism and Brain-inspired Computing Integrating DIKWP Model, https://www.researchgate.net/publication/392166001_hezuozhengjirongheDIKWPmoxingdejushenzhinengshenjingjizhiyuleinaojisuanyanjiu

Guangxi Builds Embodied Intelligence Industry-Education Integration Innovation Center to Cultivate ASEAN Talents - China Reform Forum, http://www.chinareform.org.cn/2025/0917/44515.shtml

China-ASEAN Artificial Intelligence Innovation Cooperation Training Class Started in Nanning - Guangxi Channel - People's Daily Online, http://gx.people.com.cn/n2/2025/1113/c179430-41409974.html

Draft Report on "Large Models and High-quality Datasets under Sovereign AI"- Yucong Duan's Blog - ScienceNet, https://wap.sciencenet.cn/blog-3429562-1490270.html

Frontier Review of Cybersecurity Economics: Integrated Application of DIKWP Model, Sovereign AI and Semantic Sovereignty, http://www.bilibili.com/read/cv42798380/

人工意识与人类意识

人工意识日记

玩透DeepSeek：认知解构+技术解析+实践落地

人工意识概论：以DIKWP模型剖析智能差异，借“BUG”理论揭示意识局限

人工智能通识 2025新版 段玉聪 朱绵茂 编著 党建读物出版社

主动医学概论 初级版

世界人工意识大会主席 | 段玉聪

邮箱|duanyucong@hotmail.com

世界人工意识科学院

邮箱 | contact@waac.ac