Research on the Current State of AIGC Transparency Labelling in Mainstream Media News: A Content Analysis of Five Major Platforms

Abstract

AI-generated content (AIGC) has become deeply embedded in news production, and how it is labelled is directly related to news transparency and public trust. Taking five mainstream Chinese media outlets—XinhuaNet, Chinanews.com, CNR.cn, GMW.cn, and Chuanguan News—as research objects, this study coded 153 AIGC news products through content analysis; the label position, disclosure depth, label form, and salience were examined, and chi-square tests were conducted using the implementation of the Measures for the Labelling of AI-Generated Synthetic Content (effective September 1, 2025) as the dividing point. All the samples were independently coded by two coders, and reliability testing confirmed the consistency of the coding scheme. The results show that the labelling compliance rate of mainstream media reached 100%, with label positions highly concentrated in title prefixes. After the Measures took effect, dual labelling combining “text plus visual watermark/corner mark” increased significantly (p < 0.05), reflecting improved formal standardization. However, disclosure depth did not improve in parallel, and labelling practices diverged significantly across platforms: platforms with dedicated AIGC columns and unified attribution norms achieved more standardized and deeper disclosure. The findings indicate that the Measures have promoted the formal standardization of AIGC labelling, while substantive transparency still needs to be enhanced. On the basis of these findings, the paper proposes improvements at the institutional, organizational, and communication-strategy levels, calling for a shift from bottom-line formal compliance toward substantive transparency that genuinely supports the public's right to know.