Against the backdrop of escalating China-U.S. technological competition and the deepening strategic contest within the global semiconductor industry, the boundaries and risks associated with technical intelligence activities have become highly concerned issues by international community. In 2025, China’s Ministry of Commerce included the Canadian company Tech Insights and its related entities in the “Unreliable Entity List”, marking a significant shift in China’s countermeasures in the semiconductor sector—from primary restrictions focused on hardware and equipment level to broader controls over technical intelligence and information flows. This move has triggered extensive discussion across both industry and academia. Taking the sanctions against Tech Insights as a point of departure, this article outlines the company’s business model and development trajectory, and examines how its practices in chip reverse engineering and technical intelligence services diverged from internationally recognized commercial norms and professional ethics. The article further analyzes the underlying drivers of these practices and their adverse impacts on global semiconductor innovation ecosystems as well as industrial and supply chain security. It argues that amid intensifying great-power rivalry and the rise of techno-nationalism, technical intelligence activities—traditionally regarded as commercially neutral—are increasingly being toolized and embedded within strategic competition frameworks. Under such conditions, reliance on market mechanisms solely is insufficient to reconcile the growing tension between security and development. Focusing on the question of how technical intelligence can be used legally and in compliance with regulations in an environment of heightened international competition, the article advances targeted policy recommendations. It emphasizes the importance of accelerating the establishment of a national-level international regulatory framework for competitive intelligence that is compatible with globalization and grounded in the principle of mutual benefit. Such a framework would provide critical support for maintaining fair competition and promoting international technological cooperation, and thus holds significant practical relevance and policy value.

Strategic emerging industries serve as a critical carrier for accelerating the development of new-quality productive forces, and large-scale industrial classification of enterprises constitutes a fundamental task for formulating industrial policies. To address the issues of high cost and low efficiency in conventional manual screening, an industrial classification method based on bidirectional enhancement of text data and label semantics is proposed. First, to deal with the limited training samples and unbalanced categories, we augment the textual corpus through a suite of data-enhancement strategies to increase data diversity and rectify categories imbalance. Secondly, according to the industrial definitions formulated by industry authorities, we enrich label semantics by introducing external domain knowledge, thereby expanding the semantic coverage of the labels and improving their semantic
expression capabilities. Finally, Cross-Attention mechanism is employed to achieve deep semantic interactions between text and labels which helps better capture key semantic features in text data and strengthen the intrinsic semantic alignment between text and labels, thus improving model performance. Experimental results show that the proposed new method achieves a performance improvement of more than 3.5% compared with traditional baseline models in the application of industrial classification task of three leading industries in Shanghai.

To systematically analyze the current research landscape of open science practice, this study employs a bibliometric analysis base on the Web of Science and core databases of China National Knowledge Infrastructure (CNKI), quantitatively revealing the differences between domestic and international research. It further integrates the innovation diffusion theory and the Technology-Organization-Environment (TOE) framework to construct a “dynamic-static combination” analytical model. Key findings include:（1）Technology Dimension: The adoption of technology in research exhibits a dual characteristic of demand-oriented compatibility principle (gradual integration of existing technologies) and pragmatism (emphasis on immediate effectiveness), driven by both “top-down” (government governance) and “bottom-up” (public participation) forces.（2）Organization Dimension: The analysis primarily centers on three key entities. The state/government assumes the core role in governance and policy implementation; research institutions/personnel serve as the main actors fostering practice and culture cultivation; society/the public act as deep participants, promoting scientific democratization through mutual empowerment.（3）Environment Dimension: Risks involved in the research are systematically distributed across the physical layer (technological ethics, infrastructures and talents), network layer (international competition, resource allocation), application layer (incentive-evaluation contradictions, personal concerns), and policy layer (lack of top-level design). The environmental representation manifests as the evolution of the research paradigm towards open innovation research (emphasizing transparency, sharing, and collaboration), and the reshaping
of collaborative models towards an open innovation ecosystem (characterized by co-opetition and symbiosis among multi-dimensional actors).

Recognized as a core driver for enhancing scientific and technological productivity, artificial intelligence (AI) has been incorporated into China’s major national strategies. Both central and local governments have rolled out supporting policies, propelling the “AI+” initiative to integrate AI across diverse fields. This study analyzes a sample of 450 central and local AI policies from the PKULAW database and official releases (2017-2024). By employing the BERTopic model and the policy modeling consistency (PMC) index model, it comprehensively evaluates the characteristics, thematic focus, and efficacy of these policies across three dimensions:
textometrics, content mining, and policy evaluation, thereby quantifying their coordination degrees. Findings indicate regional differences and imbalances, yet reveal high consistency and effective coordination between central and local policies, with joint priorities on the research and development and application of next-generation AI technologies, education promotion, medical devices, and talent cultivation. Central policies demonstrate the highest efficacy, with the eastern regions outperforming the central and western regions. This research clarifies the degree of central-local policy coordination, provides a reference framework for other domains, and holds pivotal significance for advancing AI industrial development and supporting national strategic objectives.

Policy comparison research has widely adopted a range of intelligence analysis methods. However, existing studies often separate macro-level textual similarity analysis from micro-level analyses of policy tool and thematic structures. This separation overlooks the interconnections between analytical levels and, to some extent, limits the overall effectiveness of policy intelligence analysis. From the perspective of intelligence studies, this paper proposes an integrated methodological framework that leverages large language models (LLMs) to enable coordinated macro–micro comparative analysis. Taking China’s new energy vehicle (NEV) policy as an empirical case, the study examines the feasibility and validity of the proposed approach. The results indicate that the policy comparative texts generated by LLMs perform well in terms of readability, informational richness, and content completeness, although a certain degree of score convergence is observed in similarity assessments. The proposed method provides empirical evidence for the application of large language models in policy comparative research while offering new analytical perspectives and methodological tools to support systematic policy analysis by intelligence researchers and policymakers.

Taking the Shanghai special exhibition industry as a research sample, this study explores the evolutionary process and ecological landscape of the industry from both diachronic and synchronic perspectives, under the framework of “co-opetition.” The Shanghai special exhibition industry, which first emerged in the early 2010s, has experienced rapid expansion and differentiation before the COVID-19 pandemic, and is now entering a new stage characterized by both phenomenal explosion and structural adjustments coexist. A diverse ecological structure, built by authoritative leaders, innovative disruptors, professional operators, and ecosystem co-builders, constitutes the complex competitive and cooperative landscape of the Shanghai special exhibition industry. The study suggests that future development paths for the special exhibition industry require abandoning zero-sum thinking and shifting towards win-win cooperation, with differentiated positioning among different entities to achieve an upgraded “multi-party co-construction” model.