On 21 February 2025 the Cyberspace Administration of China published the enforcement priorities for its 2025 Qinglang campaign. Unlike earlier editions organised mainly around the Spring Festival and summer holidays, this year’s eight tasks include “rectifying the misuse of AI technology” as a standalone item, marking a shift from ad hoc action to an annual routine.
Where AI sits among the eight tasks
The 2025 priorities cover the Spring Festival online environment, false information posted by self-media, malicious marketing in short video, misuse of AI technology, online “blackmouth” attacks on enterprises, the summer online environment for minors, tipping chaos in live streaming, and the deliberate stirring of negative emotion.
The AI task is comparatively specific: it stresses both technology-side and content-side management, stronger labelling of synthetic content, and action against disinformation generated or amplified with AI, including paid posting. Pairing technology management with content management means the regulated class is no longer only those who publish content, but also those who supply the capability. See our notes on the first phase of the AI misuse campaign and on self-media AI labelling.
Why labelling keeps coming up
Among the many points at which AI can be misused, labelling is the cheapest and most verifiable. Once synthetic content carries an identifiable label, downstream distribution, attribution and platform action all gain a handle. That is why the labelling obligation recurs across documents.
For companies that provide generation capability, the requirement has to reach the product pipeline: whether a label survives generation, editing, export and sharing, and whether it remains traceable after re-processing. See our notes on the labelling measures in force and on user verification duties for generated content.
From annual priorities to day-to-day compliance
The value of an annual priority list is that it sets the order of regulatory attention. Companies can use it to sequence compliance spending: deal with labelling and disinformation first, since they are frequent and easy to trigger, and address deeper technology-management issues afterwards.
It is worth noting that the AI theme was later expanded into a separate campaign, showing that the granularity of AI governance keeps increasing. See our notes on the dedicated AI application campaign.
For AI companies, treating an annual priority list as a one-off inspection exercise is inefficient. A better approach is to translate it into an internal content-safety checklist and a pre-publication review step, so that compliance settles into the product process.
This article is based on the Cyberspace Administration of China’s 2025 Qinglang campaign notice and public information and is for reference only; it does not constitute legal advice.