In July 2025 the Cyberspace Administration of China launched the Qinglang 2025 Summer Campaign on Minors’ Online Environment, running for two months. Unlike previous years, this campaign writes the use of AI features by minors into its scope.
Four enforcement priorities
The campaign focuses on four prominent problems: online harm directed at minors; the covert spread of illegal and harmful information through new vehicles such as trading cards and animation; luring minors into dangerous offline activities; and profiting from the images of minors.
On the enforcement path, the campaign pays particular attention to how minors’ mode is actually used, to content safety on children’s smart devices, and to the improper use of AI features in the minors’ sphere and AI-induced overuse.
Why AI companion products are the focus
The spread of generative and anthropomorphic interaction products has let “AI companionship” penetrate minors’ groups quickly. These products carry two risks at once: content risk (outputting unsuitable material) and relationship risk (forming emotional dependence and extending usage time).
The latter deserves particular attention. Traditional content governance targets what a user has seen, whereas the problems raised by AI companionship are often that a user uses it too long and depends on it too deeply — closer to an addiction mechanism than a content violation. See our note on the measures for anthropomorphic interaction services.
Minors’ mode cannot be just a switch
Minors’ mode has long suffered from a “nominal” problem: hidden entry points, thin content pools and easy circumvention. By making actual usage a focus, the campaign signals that supervision will move from “whether it is offered” to “whether it works”.
For product teams, that means providing verifiable protective effect, not merely declaring in the product description that a minors’ mode exists. See our note on the guidance on minors’ online protection.
Where this meets the AI content-governance line
The campaign echoes other enforcement running at the same time: the Qinglang campaign against false information posted by self-media, launched in late July, likewise makes AI-generated content labelling a priority. See our note on Qinglang’s action against false self-media information.
The two lines meet at one point: AI-generated content must be both identifiable (labelling) and controllable (age-graded, time-limited, scope-limited). See our note on the Measures for Labelling AI-Generated Synthetic Content.
Where companies can start
First, assess the mechanisms for identifying and protecting minors, including age recognition, time management and content grading.
Second, set default protections for AI companion features, such as enabling minors’ mode by default and capping continuous usage time.
Third, build an accessible reporting and response channel, bringing minors-related risks into a fast-handling process. See our note on the reporting channel for AI application misuse.
Minors’ protection is the most rule-dense and least error-tolerant area of AI governance. The compliance floor is not “nothing goes wrong” but “being able to prove that every effort was made to protect”. See our note on the AI education ethics reference framework.
This article is based on public information on the Cyberspace Administration of China’s Qinglang 2025 Summer Campaign on Minors’ Online Environment and is for reference only; it does not constitute legal advice.