China orders mandatory short‑video labels — creators must self‑classify, platforms enforce

.

Lede

Beijing’s Cyberspace Administration (CAC) has issued guidance making labeling of short videos mandatory at posting and retroactive tagging compulsory for existing content, citing concerns about unlabeled fiction, staged marketing and AI‑generated material. Platforms report large‑scale removals and mass relabeling as they rush to comply.

What the rule requires (plain terms)

• Mandatory tags at upload: Creators must mark whether a short video is fictional, staged marketing, or contains AI‑generated elements before publishing.
• Retroactive review: Platforms must scan and add labels to existing short videos and remove content that violates rules. Officials say platforms removed >37,000 videos, penalized >3,400 accounts, and added labels to >600,000 videos in the first enforcement period.

Why Beijing says it did this

The CAC frames the move as a consumer‑protection and misinformation measure: inconsistent labeling, it says, has allowed AI‑generated or staged clips to mislead viewers. The policy builds on earlier AI‑labelling measures introduced in 2025 that require visible tags and embedded metadata for synthetic media.

Practical effects and risks (what readers should know)

• Responsibility shifts to users. Creators now self‑classify; honest mistakes can trigger penalties or reduced reach. This increases legal exposure for ordinary users.
• Ambiguity for mixed content. Many short clips combine real footage, editing, and AI effects; creators report confusion about how to label hybrid material, encouraging over‑cautious removal or non‑posting.
• Censorship and chilling effects. Label categories like “staged” can be applied to citizen reporting, petitions, or social‑grievance videos, giving platforms a pretext to suppress content that diverges from official narratives. Rights groups document a broader pattern of tightened online controls in China.

Why this matters beyond China

The rule illustrates how labelling regimes can be repurposed for content control when enforcement rests with platforms and creators rather than independent review. European and German regulators watching platform liability and transparency debates should note how labeling requirements interact with censorship incentives.

.

.

.

Sources

• Xinhua / CAC announcement summary, March 21, 2026.
• China Daily reporting on platform removals and retroactive labels.
• The Star summary of CAC guidance.
• Legal/industry analysis: Bird & Bird on China’s AI labelling measures (2025).
• Coverage of platform compliance and AI labelling rollout (SCMP summary via Yahoo).
• Context on China’s broader censorship and digital‑rights environment (Human Rights Watch).

.