---
title: "China's Online Performers Are Losing Work to Their Own AI…"
canonical: "https://www.metaintro.com/blog/china-performers-ai-doubles-likeness-licensing-2026"
language: "en"
author: "drashtigarach"
published: "2026-09-01T13:20:20.000Z"
modified: "2026-09-28T18:21:53.917Z"
---

[Back to Blog](/blog)
[China](/blog/tag/china)[AI](/blog/tag/ai)[Media](/blog/tag/media)[Jobs](/blog/tag/jobs)[News](/blog/tag/news)
# China's Online Performers Are Losing Work to Their Own AI Doubles

95% of China's Q1 micro-dramas were AI-generated and a minute of video now costs $90. How performers responded, and what likeness licensing means for any career.

[![Drashti Garach](https://cdn.metaintro.com/rs:fill:40:40/q:72/plain/images/5719d740-e510-42bc-8017-e040d145f35f_1766029465094.png)Drashti Garach @DrashtiGarach](/blog/author/drashtigarach)

[September 1, 2026](/blog/archive/2026/09)13 min read

![China's Online Performers Are Losing Work to Their Own AI Doubles](https://cdn.metaintro.com/rs:fill:1200:675/q:78/plain/images/10112-hero-v2.jpg)

[https://x.com/intent/tweet?text=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026](https://x.com/intent/tweet?text=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026&title=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026&title=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles)[mailto:?subject=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026](mailto:?subject=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026)

The fastest displacement of a named occupation by AI anywhere in the world is happening in Chinese short form drama, and the numbers are not close. Reporting by [The Decoder](https://the-decoder.com/ai-generated-videos-are-already-displacing-actors-and-livestreamers-across-chinas-entertainment-industry/), drawing on the Financial Times, records that about 128,000 short dramas were published in China in the first quarter of 2026, three times the total for all of 2025, and that 95 percent were AI generated according to the China Netcasting Services Association. This is not a forecast about the future of work. It is a completed transition in one segment, and the way workers responded to it is the part worth studying. At [Metaintro](/), we cover displacement stories like this because the mechanics travel even when the industry does not.

## How fast did this actually happen?

Fast enough that the volume figures are hard to absorb. Production tripled the entire previous year's output in a single quarter, and almost all of the new supply was synthetic.

The trigger was a capability jump. Since the launch of a major Chinese video generation model, digital performers have been able to produce better videos faster and more cheaply than humans, and quality has jumped again with subsequent releases. Two things happened at once, which is what made the shift so abrupt. The output got good enough, and the cost collapsed.

The format was also unusually exposed. Chinese micro dramas run one to ten minutes per episode, are produced and distributed almost entirely on mobile, have a production cycle measured in days, and run on budgets that are a fraction of traditional drama series, as [China Innovation Watch](https://www.ciw.news/p/china-actors-ai-licensing) describes. A format built for speed and low cost is the first place a cheaper, faster method wins.

## What are the actual economics?

Stark, and they explain everything else. One minute of AI video now costs 90 to 120 dollars, about ten percent of what production with human actors used to cost, according to a Tsinghua University professor speaking to the Financial Times.

A ninety percent cost reduction is not a margin improvement, it is a different business. No amount of goodwill toward human performers survives that gap at scale in a format competing purely on volume and speed.

There is a counterweight in the data worth noting, because it complicates the simple story. Live action drew twenty five times the playback on one fiftieth of the launches over one major holiday period. In other words, human made content was vastly outperforming per title even as synthetic titles flooded the catalogue. That is the shape of a market producing enormous quantities of cheap content that few people watch, while the smaller human made slate does the actual work of attracting audiences.

## How many people's jobs are involved?

More than the drama industry alone. The sector employs 690,000 people directly, and 15 million people list livestreaming as their primary job.

That second figure is the one to hold onto. The displacement is not confined to actors in scripted productions, it reaches influencers and livestreamers, which is a far larger and less protected population, and one entering the same labour market as [China's college graduates](/blog/china-college-graduates-job-market). Livestreaming in particular has been a significant route into paid work for people without formal credentials, and it is now exposed to exactly the same substitution.

The employment consequences are already reaching the legal system, with AI related labour disputes rising over the past two to three years according to lawyers. The courts are already engaged, with [a Chinese court ruling AI driven layoffs illegal](/blog/china-court-ai-layoffs-illegal-ruling) and [Chinese labour law being tested by AI layoffs](/blog/china-ai-layoffs-labor-law-2026) at the same time.

## What is the practice performers are objecting to?

Being required to hand over the raw material first. Some performers are being forced to distil their voice and likeness into AI tools before being let go.

That sequence is the thing to notice, because it is the same pattern appearing in other industries under a different name. The worker supplies the input that enables the substitution, then the substitution removes the worker. The white collar version of this is [companies asking workers to train the AI that replaces them](/blog/companies-asking-workers-train-ai-replace-them-2026) and in [the low paid jobs training the AI that could replace them](/blog/low-paid-jobs-training-ai-that-could-replace-them). The Chinese performance industry is simply the most literal expression of it, because what gets captured is a person's actual face and voice.

It also produced a public dispute over consent, the same question now reaching Western hiring as [AI deepfakes enter the recruitment process](/blog/ai-deepfakes-hiring-process-2026-job-seekers). When a major streaming platform announced an AI performer library at its annual conference in April, presenting a slide naming 117 artists, several studios denied it within hours, with one stating it had not signed any AI related authorisation and that legal counsel was handling the matter, followed by parallel denials from two more. The company later retracted the claim. The chief executive had told the audience that AI could take an actor from four productions a year to fourteen, and predicted live action filming might one day become intangible cultural heritage.

## What does likeness licensing actually change?

It converts a person into an asset that can be rented, which is a genuinely new employment category rather than a variation on an old one.

The distinction matters legally and practically. Traditional likeness licensing covers completed footage that has already been shot. AI likeness licensing is different in kind, because it covers facial feature data that an algorithm uses to generate new images indefinitely. Under Chinese law both are treated under the same framework, with Civil Code articles 1018 to 1023 establishing likeness rights and extending protection to voice.

A digital twin performs scripted actions, appears in multiple productions at once and generates revenue, while the human original does nothing. Read that carefully as a description of a labour arrangement. The income is decoupled from the work, which is either the best or the worst thing that can happen to a performer depending entirely on the terms of the contract.

The terms are where the real negotiation is now happening. One licence granted by a well known actor restricts use to a single game intellectual property, image only, with no voice. A serving mainland star became the first to license an official AI twin, opening standardised multi scenario licensing in August, with the deal privately addressing several concerns the industry's performers' body had raised.

## Why does the consent dispute matter more than the technology?

Because it established that the terms are contestable, and that is the only leverage workers in this position have.

The April episode is instructive precisely because it went wrong publicly. A platform presented a slide naming 117 artists as part of an AI performer library, and within the same afternoon studios began denying they had authorised anything. The claim was subsequently retracted. Nothing about the underlying technology changed that day. What changed was the demonstration that performers would contest an assertion of consent, and that doing so quickly and collectively worked.

That is worth separating from the question of whether AI performers are good enough, which is settled and no longer interesting. The live question in every affected industry is who is deemed to have agreed to what, and on what terms. The chief executive's framing that AI could take an actor from four productions a year to fourteen is a genuine value proposition if the actor captures the upside. It is something else entirely if the studio does.

The individual deals now emerging suggest the negotiation is producing real constraints rather than blanket permissions. One licence is limited to a single game intellectual property, covers image only and explicitly excludes voice. Another established standardised multi scenario licensing. Those are the beginnings of a contractual vocabulary that did not exist a year ago, and they exist because the first attempt to proceed without one was challenged.

## How does this compare with what is happening elsewhere?

The same fight is running in several markets at once, though at different speeds and with different tools.

In video games, performers have taken industrial action specifically over AI, from [the actors' strike over AI threats](/blog/video-game-actors-strike-ai-threats) to [the industry's attempts to protect performers](/blog/protecting-video-game-performers-from-ai-threats). In broadcasting, [a media company cut jobs and replaced local news anchors with AI](/blog/scripps-268-jobs-cut-ai-anchors-local-news-2026), which is the same substitution applied to a different on screen role. In music, the response has been legal and institutional, from [labels suing AI music start ups](/blog/labels-sue-ai-music-start-ups) to [Australia barring fully AI made songs from its charts](/blog/australia-aria-ai-songs-chart-ban-2026).

What distinguishes the Chinese case is scale and completeness. Elsewhere the argument is about what might happen. In Chinese micro drama, 95 percent of a quarter's output was already synthetic, and the negotiation is therefore about the terms of a transition that has already occurred rather than about preventing one.

That makes it the most useful case study available, including for workers with no connection to entertainment. The sequence ran from capability jump, to cost collapse, to volume flood, to disputed consent, to individually negotiated terms. Any occupation whose output can be captured and regenerated is somewhere on that same path, and knowing the order of events is genuinely useful for judging where your own industry currently sits. The same repricing is reaching [average creative careers](/blog/ai-creativity-average-creative-careers-2026), and [Cate Blanchett's concerns about AI in entertainment](/blog/cate-blanchett-ai-concerns-entertainment) covers how the same questions are being raised in Western production.

## Does the same exposure exist outside China?

It does, and the US numbers show why the licensing question matters so much. The [Bureau of Labor Statistics](https://www.bls.gov/ooh/entertainment-and-sports/actors.htm) puts the median hourly wage for actors at 29.05 dollars in May 2025, with employment projected to show little or no change from 2025 to 2035. A workforce paid by the hour, with flat projected demand, has very little slack to absorb a technology that reproduces its output at near zero marginal cost.

The regulatory answer is arriving from several directions at once. In China, the [Cyberspace Administration](https://www.cac.gov.cn/) sets the rules for synthetic media and labelling, and the [State Council](https://english.www.gov.cn/) publishes the national policy those rules sit under. Internationally, likeness and performance rights fall under the frameworks maintained by the [World Intellectual Property Organization](https://www.wipo.int/). The [United Nations cultural agency](https://www.unesco.org/) has published its own recommendation on the ethics of artificial intelligence, which covers cultural and creative work explicitly. None of those bodies decides an individual contract, but they set the boundary any licence has to be written inside, which is why the terms being negotiated now will outlast this particular news cycle.

## What does this mean for your career?

Five points, and they apply well outside entertainment.

Understand that the response was terms, not refusal. The most useful framing here is that China's actors did not reject AI, they demanded terms. Refusal is rarely available to an individual worker facing a ninety percent cost advantage. Conditions sometimes are, and the difference between a licence restricted to one product with no voice and an open ended grant is the difference between an asset and a giveaway.

Read anything you sign about your voice, face or recorded output very carefully. The critical distinction is between licensing completed work and licensing the data used to generate new work indefinitely. Those look similar in a contract and are not remotely the same. If your industry is starting to ask for the second, that is the moment to get advice.

Assume the pattern reaches you even if your job has no camera in it. The mechanism is capture, then substitution, and the raw material can be process data as easily as a face. The nearest Western equivalents are [AI video tools reshaping entertainment work](/blog/ai-video-tools-creative-jobs-entertainment-workers-2026) and [AI splitting creative careers across design and film](/blog/ai-splitting-creative-careers-design-film-jobs-2026), alongside the disputes in [video game voice actor strikes over AI](/blog/videogame-voice-actors-strike) and [protecting video game performers from AI threats](/blog/protecting-video-game-performers-from-ai-threats) show the same negotiation running elsewhere.

Watch where the audience actually goes rather than where the volume goes. Human made live action was drawing twenty five times the playback on a fraction of the launches. Volume and value are separating, and the durable work tends to follow the second, which is also what [Adobe's AI agents mean for creative work](/blog/adobe-ai-agents-creative-work-2026) once the novelty wears off.

Finally, treat collective bargaining as the realistic route to decent terms. The individual licences being negotiated came after a performers' body raised concerns publicly, and the platform retraction followed studios speaking up together within hours. That is a faster and more effective correction than any individual could have achieved alone. The same lesson is playing out in another market, where [voice actors face AI dubbing displacement](/blog/voice-actors-ai-dubbing-workforce-displacement-2026).

## Related Articles

- [China Is Quietly Bracing for an AI Jobs Shock](/blog/china-bracing-ai-jobs-shock-2026)
- [Stealth Layoffs Hit China's Tech Giants as Managers Push Workers to Quit](/blog/china-tech-stealth-layoffs-2026)
- [Disney Brings AI Into Its Parks and What It Means for Creative Jobs](/blog/disney-ai-parks-creative-jobs)
- [AI Scraping Is the Newest Media Industry and It Is Replacing Reporters](/blog/ai-scraping-media-industry-2026-replacing-reporters)
- [How to Land a Content Creator Job](/blog/content-creator-job)
- [China's Dark Factories Are Eliminating Jobs](/blog/china-dark-factories-ai-robotics-eliminating-jobs-2026)
- [Marvel Studios Loses Its Visual DNA and What Disney's Cuts Mean for Creatives](/blog/disney-damaro-marvel-studios-layoffs-creative-workforce-2026)
- [What China's Midlife Layoff Crisis Reveals About Age Bias](/blog/china-midlife-layoffs-global-age-bias-labor-market)
- [The BBC's Biggest Workforce Reduction in Years and What It Means for Media Jobs](/blog/bbc-2000-job-cuts-2026-media-layoffs)
- [China as the Emerging Superpower of the Global Labour Market](/blog/china-superpower-labor-market)
- [When a Candidate's Social Media Costs Them the Job](/blog/candidate-social-media-costs-job-protect-profile)
- [Comcast Spins Off NBCUniversal and What It Means for Media Jobs](/blog/comcast-nbcuniversal-spinoff-media-jobs-2026)

## People Also Asked

### Q: How much of China's short drama output is now AI generated?

A: Almost all of the new supply. About 128,000 short dramas were published in China in the first quarter of 2026, three times the total for all of 2025, with 95 percent AI generated according to the China Netcasting Services Association. A separate analysis put AI at 95 percent of first quarter micro drama supply by title count.

### Q: Why is AI video replacing performers so quickly in this format?

A: Cost, combined with a format built for speed. One minute of AI video now costs 90 to 120 dollars, about ten percent of what production with human actors used to cost. Chinese micro dramas run one to ten minutes per episode, are produced almost entirely for mobile, have production cycles measured in days and run on small budgets.

### Q: What is a digital twin licence and how is it different from normal likeness rights?

A: It covers the data rather than the footage. Traditional likeness licensing applies to completed footage already shot, whereas AI likeness licensing covers facial feature data that an algorithm uses to generate new images indefinitely. A digital twin performs scripted actions, appears in multiple productions at once and generates revenue while the human original does nothing. Chinese Civil Code articles 1018 to 1023 establish likeness rights and extend protection to voice.

Future proof your career by understanding what you are actually signing away. [Metaintro](/) tracks how AI is reshaping pay, rights and job security across industries worldwide, not just in tech. [Create your free Metaintro profile](/signup) to follow the changes reaching your own field.

### Share this article

[https://x.com/intent/tweet?text=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026](https://x.com/intent/tweet?text=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026&title=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026&title=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles)[mailto:?subject=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026](mailto:?subject=China's%20Online%20Performers%20Are%20Losing%20Work%20to%20Their%20Own%20AI%20Doubles&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fchina-performers-ai-doubles-likeness-licensing-2026)

For job seekers

## Skip the job boards. Keep the career.

Tell Metaintro your story once. It reads every posting, every day, and brings you only what fits.

[Get Started Free](/signup)[Search matching jobs](/jobs/search)

- MatchCompare live roles against your current evidence.
- PositionTurn proof projects into role-specific applications.
- ImproveUse market feedback to keep the skill plan current.

[Return to navigation](#main-navigation)