The ethical principles governing scientific data publishing within academic publishing institutions are investigated, aiming to clarify their conceptual foundations and practical elements while offering guidance for future normative construction in China. Ten leading international publishing institutions with extensive experience in scientific data publishing were selected, and a total of ten representative policies were collected and analyzed. The policies were retrieved through systematic web-based investigation, and content analysis was utilized to code and interpret the ethical elements embedded in the policy texts. Using the scientific data publishing workflow—encompassing submission, review, storage, publication, and citation—as the analytical framework, this study identified both overarching and specific ethical principles and further distilled their operational implications. The analysis revealed a two-tiered framework comprising two basic principles and seven specific principles. The basic principles—value co-creation and security—function as the overarching value orientation and risk-control baseline across the entire publishing process. The seven specific principles correspond to distinct workflow stages: transparency, data ownership protection, and privacy protection (submission); quality control (review); optimal matching (storage); open licensing (publication); and standardized citation (citation). Each principle was carefully examined through qualitative coding of policy statements, supplemented by comparative analysis across institutions, enabling the clarification of conceptual boundaries and the extraction of actionable elements. For instance, the principle of transparency is implemented primarily through mandatory data availability statements; privacy protection relies on obtaining informed consent and applying data anonymization techniques; quality control necessitates collaborative efforts among authors, editors, and reviewers to safeguard data integrity; and the principle of optimal matching emphasizes disciplinary alignment and the selection of high-quality repositories. Similarly, the principle of open licensing is enforced by encouraging authors to adopt Creative Commons agreements such as CC0 or CC-BY, thereby lowering barriers to reuse; the principle of standardized citation obliges authors to include complete dataset information and persistent identifiers (e.g., DOIs) in reference lists to ensure traceability and proper recognition of scholarly contributions. This study determines that leading international academic publishing institutions have accumulated mature experience regarding scientific data publishing ethics, establishing ethical principles that cover the entire data publishing process. Given that scientific data publishing in China remains in its early developmental stages and lacks systematic experience in building ethical norms, this paper proposes four feasible pathways: (1) strengthening publicity and institutional guidance to increase researcher participation; (2) formulating operable ethical guidelines and embedding them into the data publishing workflow; (3) strictly adhering to security boundaries by constructing a “self-check plus external review” mechanism for data security governance; and (4) leveraging the functions of multiple stakeholders to promote effective implementation of data review mechanisms. By systematically analyzing policy texts and integrating insights into an ethics framework, this study contributes both theoretically and practically to understanding scientific data publishing ethics. It provides academic publishing institutions in China with a concrete reference for constructing localized, operable ethical norms that balance openness with protection, thus facilitating the healthy development of scientific data publishing.