China Says 430,000 Microdramas Arrived in Eight Months—More Than 90% Used AI
ReelPulse Editorial
Sep 21, 2026
The regulator's figure is a production count, not an audience ranking
China's State Council Information Office reported on September 20 that an official from the National Radio and Television Administration put January-through-August 2026 microdrama output at approximately 430,000 titles—13 times the total reported for the previous year. The official said more than 90% were produced using AI technologies.
Those numbers describe output in China under the regulator's category definitions. They are not global totals, app-store revenue, completed premium series, unique intellectual properties or proof that viewers watched every release. The briefing does not publish the underlying title list or define the minimum amount of AI use needed for a production to enter the AI share. ReelPulse therefore treats the figures as an attributed regulatory snapshot rather than an independently audited census.
Scale now means more than app revenue
The same briefing says microdramas have reached more than 800 million viewers in China and that the domestic market exceeded RMB 100 billion—about US$14.97 billion—in 2025. Both figures were attributed to the NRTA official.
The RMB 100 billion figure covers a domestic industry category and should not be compared directly with mobile in-app purchase estimates. Viewer reach, paying users, ad-supported viewing, production services, licensing, tourism tie-ins and commerce are different measures. The useful signal is structural: Chinese policy now treats microdrama as an emerging cultural business, while the format is being connected to tourism, sports, e-commerce and regional storytelling through the regulator's “micro drama plus” program.
The output boom is being met with enforcement
China SCIO says authorities removed 68,000 microdramas and directed platforms to act against more than 1,200 accounts since the start of 2026. More than 90% of the content targeted in those actions involved AI-produced microdramas, according to the official. The cited concerns include harmful or vulgar material, content considered inappropriate for children and copyright violations.
That 90% enforcement share is not an AI failure rate. It cannot be divided into the 430,000 production count without confirming identical date ranges, platforms, units and denominators. At most, the two figures show why automated supply and automated review cannot be treated as a complete quality-control system.
Disclosure and permission are becoming production requirements
The regulator's stated direction is concrete: AI-produced dramas must be labeled; platforms should verify permission for a person's likeness or voice; and checks should cover production and distribution. At the same time, platforms are being encouraged to create dedicated live-action sections and provide financial support for well-produced titles.
This is a dual policy, not a blanket rejection of AI. The state is supporting AI-assisted efficiency while drawing a firmer line around provenance, consent, content quality and rights. A production that cannot identify its references, voice permissions, model outputs and review decisions will have a harder time demonstrating compliance than a team that logs those fields from the first shot.
For global producers, evidence becomes part of the deliverable
China's rules do not automatically govern every production or distributor worldwide. The broader operational lesson still travels: faster generation increases the number of assets that need origin records, consent checks, continuity review and disclosure decisions.
ReelPulse recommends keeping a scene-level ledger with the model and version, prompts and references, voice and likeness permissions, rejected attempts, human edits, final approver and publication label. Cost should be measured per usable finished second after rejected generations—not only per generated second. This week's data makes the tradeoff visible: supply can grow extremely quickly while accepted, rights-cleared and audience-worthy stories remain scarce.
Primary source