China's top broadcast regulator has introduced landmark rules for the country's booming digital entertainment market, bringing AI-generated short dramas under unified national oversight for the first time alongside live-action short dramas.
The new policy measures on development and management of micro short dramas, enacted on Sept 1 by the National Radio and Television Administration, subject titles generated using artificial intelligence to mandatory pre-release filings, episode-by-episode AI labels, and strict tiered administrative reviews. The move formalizes an ecosystem that previously hovered in a regulatory gray zone between casual short videos and full-scale series.
The regulatory crackdown follows an explosive surge in AI-generated production volume paired with widespread viewer backlash to a market plagued by disorder.
Micro short dramas — ultra-concise serials with episodes typically lasting only one to two minutes — have dominated mobile screens across China in recent years. Data from the China Network Audiovisual Association shows that media platforms released over 367,000 series of micro short dramas in the first half of 2026 alone.
AI-generated series — dramas produced entirely or primarily using generative AI software rather than live camera crews — accounted for more than 74 percent of that volume. On average, over 1,500 new AI series are published on mobile platforms daily. According to market research firm DataEye, the overall user base for AI-assisted short series skyrocketed from 120 million in 2025 to over 600 million in the first half of 2026.
On short-video platform Douyin, the Chinese version of TikTok, 221,900 AI drama series were published in the first half of the year. Of them, only 1,055 exceeded 100 million views, meaning that one hit is produced for every 210 series released.
However, rapid scale has brought severe quality control problems. Viewers have complained of repetitive tropes, such as domineering executive romances, time-travel themes, and instant revenge stories — alongside instances of clickbait, copyright theft and unauthorized digital likeness clones.
In March, model Qi Hai discovered that the popular AI serial Taohuazan on the short drama app Hongguo featured an AI character identical to her, down to a distinctive mole on her face. The series, which garnered 40 million views, used her image without consent.
Another high-profile AI production was reported to have modified its male protagonist only after the legal team for actor Luo Yunxi objected to a near-identical digital replica of his likeness.
He Biao, deputy director of the National Radio and Television Administration, said at a micro short drama forum in early November that the faster AI short dramas develop, the more important it is to ensure content quality. The key to ensuring content quality lies in standardized governance, he said, stressing the need to regulate and guide the development of AI short dramas.
"Regulating AI short dramas does not mean not developing them. It means stabilizing volume and improving quality. The boundaries of AI, the copyright bottom line and the ethical bottom line must be held," he said.
Lin Qiwen, vice-president of DataEye, said the implementation of the measures fills a gap in industry rules and gives AI short dramas a clear identity — they no longer hover ambiguously between short videos and long drama series, but are formally brought into the unified micro short drama regulatory framework.
"Compared with the scattered rectification rules of the past two years, this regulation is ex ante access and whole-chain, regularized and institutionalized supervision of AI short dramas," Lin said.
He added that the explosive growth of AI short drama production capacity, together with accumulated problems such as content homogenization, uneven quality and infringement, has made clear rules necessary for the industry and laid a foundation for AI short dramas to move toward quality-driven development.
To combat digital deception, the NRTA measures enforce a pioneering requirement that every episode of an AI-generated micro drama must display a clear, prominent prompt informing viewers of its synthetic origin.
"Clear labeling prioritizes protecting viewers' right to know and reducing the risk of them being misled by false realistic content. It also forces producers to take the initiative in controlling AI content quality, and makes regulatory verification more intuitive," Lin said.
While major streaming platforms have begun adding mandatory tags at the start of episodes, smaller studios have attempted to bypass the rule by shrinking disclaimers into screen corners to maintain "viewer immersion" — a practice regulators are actively working to eliminate.
Tiered system
Beyond transparency labels, the measures address the unique economics of generative AI through a calibrated, three-tiered administrative system. Because AI production slashes production budgets from hundreds of thousands of yuan down to a fraction of traditional costs, regulators established financial review thresholds specifically tailored to AI content.
High-budget AI productions — defined as works with an investment of 800,000 yuan ($120,000) or more, as well as any series covering sensitive themes such as politics, national security, military, or religion — face the strictest oversight. These projects must undergo national NRTA pre-filing, full review of the finished work, and receive an official distribution license before release.
Mid-tier productions, with budgets between 300,000 and 800,000 yuan, are subject to review and registration by provincial-level broadcast authorities. Meanwhile, low-cost productions under 300,000 yuan can be reviewed internally by the hosting platform, assigned a program registration number, and launched directly.
These budget triggers reflect a sharp regulatory pivot designed for synthetic media. Traditional live-action micro dramas typically require budgets of 2 million to 3 million yuan before triggering top-level licensing. By lowering the threshold to 800,000 yuan for AI, regulators aim to prevent creators from mass-producing cheap, high-risk content that bypasses national oversight based on its "low-cost, fast output" characteristics, Lin explained.
Human element
At a micro short drama seminar held by the NRTA in late September, Zhang Lanshan, a professor at Beijing University of Posts and Telecommunications, said that AI can make narratives more industrialized and standardized.
"A human-led zone must be established in the core links of narrative, such as the values of creation, story themes and character emotions. At the operational level, we should promote 'human first, machine second'. AI is responsible for expanding the boundaries of creation, while humans are responsible for guarding the soul of narrative," Zhang said.
She added that when AI has the ability to produce works independently and through the full process, it will further force human creators to shift from competition in efficiency to competition in value, so that quality-driven creation ultimately returns to narrative depth, unique aesthetics and cultural value.
This also explains why neither labeling nor classification bans AI. The rules do not require AI dramas to "mark where AI is used", only to mark that "it is AI-generated"; nor does the classification shut low-cost creation out, Zhang said.
Implementation challenges remain. Regulators are still refining operational definitions, such as the exact percentage of AI assets that triggers mandatory labeling, and how self-review standards should be unified across competing streaming platforms.
To enforce compliance, on Nov 6 the NRTA launched a two-month enforcement campaign targeting low-brow themes, copyright violations and technical loopholes. As the market adapts, the regulatory message is clear: AI will be welcomed as a production tool, but not as an unchecked substitute for human creativity and legal accountability.
dengzhangyu@chinadaily.com.cn
