RedNote, known in China as Xiaohongshu, one of the country’s most influential lifestyle and social commerce platforms, has set out a clearer framework for governing AI-generated and AI-assisted content, following a series of platform notices published between February and June 2026.
The notices show a broader move by Chinese social platforms towards stronger AI content labelling, tighter control of synthetic media abuse and regularised moderation of AI-generated content that may mislead users, infringe rights or affect minors.
On 12 February 2026, RedNote published the “Announcement on Strengthening Voluntary Labelling of AI-Generated and Synthetic Content”. The platform said it had received multiple user reports involving AI-generated synthetic technology being used to imitate public figures and produce false videos. RedNote said such unauthorised content may infringe the lawful rights and interests of public figures and undermine community trust when AI-generated or synthetic content is not clearly labelled.
The platform urged creators to proactively label AI-generated or synthetic images, text and videos, stating that disclosure of AI creation methods does not reduce the value of creative work, but makes communication more transparent. RedNote also said it would continue improving its detection capability under China’s Measures for the Labelling of Artificial Intelligence-Generated and Synthetic Content. Where creators fail to actively declare AI-generated or synthetic content, the platform may add labels based on detection results, and unlabelled content may face restricted distribution.
On 15 May 2026, RedNote published the “Clear and Bright: Special Announcement on Rectifying Disorder in AI Applications”. The platform said it was responding to relevant work arrangements from the Cyberspace Administration of China and participating in the Shanghai Cyberspace Administration’s special campaign to rectify disorder in AI applications.
The notice stated that AI content labelling is both a platform requirement and a legal obligation. RedNote said AI-generated images, text, audio and video; AI-retouched, AI-edited or AI-dubbed re-creations; and AI face-swapping or voice-changing content should be labelled as “AI-generated” or “AI-assisted creation” through the platform’s publishing interface.
RedNote also identified several prohibited behaviours, including deliberately concealing AI-generated attributes, failing to label AI content, using labels improperly, removing or damaging existing AI labels, spreading tutorials or tools for removing AI labels, and providing or selling AI label-removal services.
The platform said it would strictly act against AI fully automated managed accounts, bot accounts with no real human participation, AI impersonation, fabricated identities, false information, rumour creation, AI infringement, unauthorised face-swapping or voice-swapping, infringement of portrait rights and copyright, and low-quality mass-produced AI content. RedNote said that since the start of the year it had handled more than one million cases of harmful AI-related behaviour.
On 1 June 2026, RedNote published its eighth governance notice on “AI-modified” videos, focusing on AI-generated audiovisual content that distorts classic works, historical subjects, revolutionary themes or well-known animation images. The platform said such content may affect public understanding, erode traditional culture and interfere with minors’ formation of correct cultural understanding and perception of reality.
RedNote said it had incorporated “AI-modified” videos into regular governance, using stronger technical identification, expanded strategy keyword databases, refined control priorities and multi-stage clean-up across notes, comments and user accounts.
The platform said that in May 2026 it newly cleaned up and handled 994 pieces of violating “AI-modified” video content and took action against 17 violating accounts. It cited examples including user Xiao6, who was involved in frequently publishing AI-modified content relating to historical figures; user Jing, who was involved in AI-modified content relating to Dream of the Red Chamber; and user 00**le, who was involved in AI-modified content relating to Journey to the West. The accounts were muted by the platform.
“The three notices should be read together as a clear signal that AI content governance in China is becoming more systematic,” said Jone Smiths, spokesperson for Gogetop Marketing, a London-based marketing and communications agency working with international brands across the UK, European and Chinese markets. “The focus is moving beyond individual AI-generated posts towards platform-wide risks, including synthetic identity, misleading content, automated accounts, copyright and portrait rights, cultural distortion and youth protection.”
For brands, agencies and creators using RedNote, the notices point to a practical shift in content operations. AI may remain a useful tool for content production, localisation and creative testing, but AI-assisted content now requires clearer internal review, more careful labelling and stronger attention to platform-specific rules.
The platform’s latest notices suggest that the next phase of social media marketing in China will be shaped not only by creativity and speed, but also by transparency, compliance, authenticity and user trust.
