China's entertainment industry is undergoing an AI-driven transformation faster than almost anywhere else in the world, and the people who make its content are paying the price. According to a report by the Financial Times, AI-generated videos are already displacing actors and livestreamers across China, with entire segments of the production economy shifting to synthetic performers that are faster, cheaper, and available on demand.
The numbers behind the shift are striking. In the first quarter of 2026 alone, roughly 128,000 short dramas were published in China — three times the total for all of 2025, according to the China Netcasting Services Association. Ninety-five percent of them were AI-generated. For more context on this story, see our ongoing AI trends.
The Numbers Behind the Shift
Short dramas — vertically shot, fast-paced serials optimized for mobile viewing — have exploded into one of China's biggest digital entertainment categories. The industry directly employs about 690,000 people, and 15 million people list livestreaming as their primary job, per the association's figures cited in the FT reporting.
Those are the jobs most directly in the path of automation. What used to require actors, crews, and shooting schedules can increasingly be produced by a small team — or a single creator — using generative video models. At the volume China's short-drama platforms publish each quarter, even partial substitution translates into tens of thousands of roles that no longer need filling.
The pace is what distinguishes this from earlier automation waves in media. A tripling of output in a single quarter, with the overwhelming majority synthetic, suggests producers are not treating AI video as an experiment anymore. It is the default production method for the category's growth.
Seedance and the New Economics of Video
The turning point came with ByteDance's Seedance 2.0, the FT reported. Since its launch, digital performers can be produced with better quality, faster, and at lower cost than their human counterparts for many kinds of content. Quality has jumped again with Seedance 2.5 and Alibaba's Wan 3.0, keeping pressure on producers to automate further.
The economics are stark. One minute of AI video now costs roughly $90 to $120 — about ten percent of what production with human actors used to cost, Tsinghua University professor Shen Yang told the Financial Times. For an industry that publishes six-figure volumes of content per quarter, that difference reshapes greenlight decisions overnight.
Cost alone does not explain the shift; iteration speed does. Synthetic production compresses the gap between concept and published episode from weeks to hours, letting studios chase trending topics and algorithmic niches in near real time. In a market where discovery is driven by recommendation feeds, speed of supply has become as valuable as production value — and AI video wins on both.
Performers Under Pressure
The human cost is mounting. Some performers say they are being forced to "distill" their voice and likeness into AI tools before being let go — effectively training their own replacements, the FT reported. Lawyers say AI-related labor disputes have risen over the past two to three years as the practice spreads through the industry.
The distillation fights echo battles already familiar in Western markets, where voice actors and performers have campaigned for consent and compensation regimes around digital replicas. China's version is unfolding at greater speed and scale, and with fewer established guardrails: the combination of mature video models, a massive creator economy, and permissive production norms has made the country the first large-scale test case for synthetic entertainment labor.
Industry bodies have begun publishing data on the shift, but meaningful protections for performers remain limited. For the 15 million people who livestream for a living, the lesson arriving from the short-drama sector is that synthetic substitutes are not a future risk — they are this quarter's competition.
The Livestreaming Parallel
The displacement is not confined to scripted content. Livestreaming — the larger of the two labor pools — is being reshaped by the same tooling in a different way. Virtual hosts and synthetic presenters can run longer hours, in more languages, without contracts or rest, and the FT reporting describes digital performers competing directly for the audience attention that human livestreamers depend on.
That creates a two-front squeeze for the 15 million people who livestream for a living: on one side, synthetic presenters undercutting the economics of human-run channels; on the other, AI production tools lowering the barrier for new entrants to flood the same feeds with content. Lawyers cited in the reporting say disputes tied to these tools have been accumulating for two to three years — and are still rising.
A Preview of Global Disruption?
What is happening in Chinese short dramas and livestreaming is likely to reach other markets as the same models — or equivalents from Western labs — improve. Voice cloning, digital extras, and fully synthetic leads are already controversial in Hollywood, and China's experience offers a preview of the arguments to come: lower costs and more content on one side, displaced workers and consent fights on the other.
It also offers a preview of the pace. Regulators elsewhere have years, not decades, to settle likeness-rights and disclosure questions before the same economics arrive on their doorstep. China's short-drama sector shows how quickly a content category can flip from human-made to synthetic once the cost line crosses — measured in weeks, not union negotiation cycles.
For now, the trend line in China is unambiguous. Ninety-five percent of new short dramas are already synthetic, and the tools producing them are getting cheaper every quarter. The question for the 690,000 people the industry employs is no longer whether AI video changes their jobs — it is how much of the work is left when the transition is done.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
