China’s Labeling Regime for AI-Generated Misinformation: Implementation Challenges and Reform Priorities

Abstract

While generative artificial intelligence (generative AI) is reshaping digital content production, it also amplifies the risks and governance challenges associated with misinformation. China’s current approach centers on a labeling regime that establishes a full-chain responsibility framework across content generation, dissemination, distribution, and use, and that adopts a dual-track system of explicit and implicit labels. However, implementation challenges persist. At the level of governance subjects, key stakeholders often lack sufficient incentives to comply, and the allocation of responsibilities among actors remains unclear. Technically, labels are vulnerable to removal or tampering, undermining traceability and enforcement, while fragmented standards limit cross-border interoperability. In terms of regulatory efficacy, labeling indicates the mode of content production but does not resolve substantive questions of authenticity; moreover, generalized labeling may induce labeling fatigue and weaken the warning function over time. To address these problems, China should clarify labeling duties across the content lifecycle, strengthen anti-tampering and detection technologies, promote interoperability with international standards, and adopt a tiered and risk-based labeling framework to mitigate labeling fatigue. These reforms can support a healthier, more orderly, and sustainable digital information environment.