摘要：

目的 探讨AI在国内外学术期刊出版中的使用边界，梳理主要出版环节中AI应用的合理范围，为完善科学规范的学术期刊AI治理框架提供参考。 方法 采用文献调研法、网络研究法与内容分析法，筛选出国内外多学科核心学术期刊发布的52份AI政策，从核心权责、具体应用场景、披露与治理机制等维度展开调研。 结果 国内外学术期刊在核心治理底线方面达成基本共识，AI使用呈现明显的场景差异化特征：辅助性写作环节相对开放，图像与数据处理层面趋于严格，同行评审环节普遍限制或禁止使用。同时国际期刊政策具体使用规范较为细致，但版权判定与惩戒机制仍有待完善；国内期刊通过强化责任追究确立治理底线，在集群化治理、规范更新、披露细化以及技术治理手段多元化等维度仍存在完善空间。 结论 学术期刊出版AI治理框架可从标准统筹、权责划分、透明披露和人机协同四个方面加以完善，推动形成权责清晰、流程透明、人机协同、可追溯且可问责的学术期刊出版治理体系，促进AI在学术期刊出版领域的规范应用与健康发展。

关键词: 学术期刊出版, 人工智能使用边界, 人工智能治理

Abstract:

Purposes To explore the boundaries of artificial intelligence （AI） use in domestic and international core academic journals， clarify the appropriate scope of AI applications across major publishing stages， and provide references for improving a scientific and standardized AI governance framework for scholarly publishing. Methods Using literature review， web-based investigation， and content analysis， this study collected and analyzed 52 AI policies issued by domestic and international multidisciplinary academic journals. The policies were examined from the perspectives of core responsibilities， specific application scenarios， disclosure requirements， and governance mechanisms. Findings Domestic and international academic journals have reached broad consensus on fundamental governance principles. The use of AI demonstrates clear scenario-specific characteristics： AI-assisted writing is relatively open， while AI applications in image and data processing are subject to stricter regulation， and the use of AI in peer review is generally restricted or prohibited. International journals tend to provide more detailed operational guidelines， although challenges remain regarding copyright attribution and sanction mechanisms. Domestic journals establish governance boundaries primarily through strengthened accountability requirements， yet further improvements are needed in collaborative governance， policy updating， disclosure refinement， and the diversification of technical governance measures. Conclusions The AI governance framework for scholarly publishing can be further enhanced through four dimensions： coordinated standards， clear allocation of responsibilities， transparent disclosure， and effective human–AI collaboration. Such efforts will contribute to a governance system characterized by clear accountability， transparent processes， traceability， and responsible oversight， thereby promoting the standardized and sustainable application of AI in scholarly publishing.

Key words: Scholarly publishing, Boundaries of artificial intelligence use, Artificial intelligence governance