The Cyberspace Administration of China (CAC, 中央网信办) announced a two-month national campaign against fake personas in short videos on 30 September 2026. The notice names five types of fabricated persona and four enforcement measures, including a ban on such videos in rankings and recommendations. In this article, we look at the targets, the measures and the gaps in the notice.
The CAC campaign runs nationwide for two months
According to the notice on the CAC website, the campaign is part of the regulator’s “Qinglang” (清朗) series. Its full title is “Qinglang: rectifying the problem of fake personas in short videos” (清朗·整治短视频虚假人设乱象).
The CAC states two aims: to promote the healthy and orderly development of the short-video industry, and to create a clean online space. The notice gives the duration as two months but does not give a start or end date.
| Item | Detail |
|---|---|
| Issuing body | Cyberspace Administration of China (中央网信办) |
| Title | 清朗·整治短视频虚假人设乱象 (special campaign) |
| Document number | Not disclosed in the source |
| Published | 30 September 2026 |
| Duration | Two months, nationwide; start and end dates not stated |
| Scope | Short-video accounts, MCN agencies, website platforms and local cyberspace authorities |
Five persona types are targeted, from staged delivery riders to AI-cloned public figures
The notice lists five problems. Several involve selling: fake farmers who sell goods “in the name of helping farmers” (助农), fake fortune-tellers who sell items with “change your luck” pitches, and AI-made videos of public figures used for false endorsements and marketing.
| Problem | Behaviour named in the notice |
|---|---|
| 1. Posing as vulnerable groups | Staged scenes such as “delivering food with a child” or “highly educated delivery rider”; posing as farmers or people in hardship to sell goods; posing as disabled or seriously ill people; posing as heartbroken singles to draw traffic for profit |
| 2. Fabricated professional credentials | Invented CVs, honours and titles; posing as doctors, lawyers or teachers; using vague professional keywords to get around platform credential checks |
| 3. AI and impersonation | AI-synthesised audio or video of public figures for false endorsements and marketing; posing as people in trending events or their relatives; “shadow accounts” built from reposted content |
| 4. Extravagance and superstition | Renting luxury cars, homes and goods for staged scenes; “rich heir” personas; fake rags-to-riches stories; feng shui and astrology personas selling goods, paid readings and private-traffic diversion |
| 5. Fabricated conflict plots | Posing as charity workers; staged rescues and fake donations; invented domestic violence, kidnapping or abandonment storylines; “human nature tests” that demean regions or groups |
Accounts produced in batches through MCN agencies face stricter action
The first of four measures is stricter action against accounts and platforms. The CAC says it will act against the use of fake personas for malicious marketing and profit, including content produced in batches through MCN agencies and distributed through multi-account matrices.
Accounts of a “severe nature” (性质恶劣) are to be banned. The notice does not define the threshold for that term.
Content labels such as “AI-generated” or “marketing” cannot be used to avoid the rules
The second measure covers labelling at the point of publication. Fake-persona videos that are not labelled as required will be dealt with strictly, according to the notice.
Platforms must carry out content review. They may not use content labels such as “fictional performance”, “AI-generated”, “marketing”, “reposted” or “personal opinion”, or “no harmful intent” disclaimers, to sidestep online-ecosystem governance requirements.
Rankings and recommendations must not carry fake-persona videos
The third measure bans pushing fake-persona short videos in rankings, recommendation slots and similar positions. It also calls for stricter credential review.
Fields that lack an occupational classification, qualification standards or certification requirements, and that offer serious room for illegal profit, are to be added to routine governance priorities by category.
The fourth measure covers specific groups. The notice names personas such as the “AI domineering CEO” (AI霸总) and the “wellness master” (养生大师) used to deceive older people, as well as rumours that demean rural residents.
Scope: local regulators and platforms are told to act
- Local cyberspace authorities (各地网信部门) are told to fulfil their territorial management responsibilities.
- Website platforms are told to carry out self-inspection and correction and to check for loopholes.
- The notice covers content creation, traffic distribution and review management as the key stages.
- The notice refers to “website platforms” in general and does not name Douyin, Kuaishou, Weixin Channels, Xiaohongshu (RedNote) or any other service.
Not disclosed in the source
- A document number and the exact start and end dates of the two-month campaign
- Which platforms or accounts are in scope by name
- Penalties other than account bans, such as fines, and the threshold for a “severe nature”
- The number of accounts or videos already handled
- How brands that commission creator content are treated
