Recently, “#A large number of AI bloggers have stopped posting.It shot straight to number one on China’s Weibo hot search list. Clicking into the related topics, one after another were shutdown announcements: accounts making AI comics, those teaching AI tools, and channels built on AI-generated art all went dark at nearly the same time. Shortly after, “a wave of million-follower influencers stopping updates” also made the trending list. China’s AI content industry, which had exploded in growth over the past year, is now facing its first large-scale retreat. The suspensions are concentrated in the general knowledge, AI tool tutorial, and AI comic sectors. Multiple Chinese media outlets have summed it up as “triple pressure”: tighter regulation, high costs, and shrinking revenue—three crushing burdens hitting the exact same window. The era of easy monetization through template rehashing and traffic-chasing has come to an end.
A large number of AI bloggers have stopped updating! With the three major pressures of regulation, cost, and revenue, China’s AI content bubble has burst.
The first mountain: stricter copyright enforcement makes template-based approaches ineffective.
Regulation has landed earliest. Starting April 1 this year, China’s National Radio and Television Administration required AI animation dramas to be “filed before release,” with any unfiled content removed from all platforms. Short-drama platforms like Fanqie and Hongguo followed with new rules: AI content episodes must not exceed 15 minutes, plot excerpts must not exceed 10% of the original work, and compilation formats are banned. Under this regulation, the past practice of mass “unauthorized repurposing” in AI animation dramas has been cut off outright. The old playbook of adapting web novels and launching them without paying licensing fees has become a pariah.
More importantly, mainstream AI models are now fully embedded with invisible watermarks. Face-swapping, voice cloning, and unauthorized adaptations can all be traced back to their source, and violators’ commissions go straight to zero. Just securing adaptation rights for a web novel could take an individual blogger a whole month, pushing the barrier up to a height most people simply can’t reach.
The second mountain: computing costs are skyrocketing—a video that lasts just a few dozen seconds takes a whole day to render.
There’s always a misconception out there that AI content creation is a “zero-cost startup.” In reality, every step of this production line burns money. One creator admits that a finished AI video of just a few dozen seconds—from selecting models, writing prompts, repeatedly generating frames, to editing and fine-tuning—takes at least three to five hours, and complex content can take a full day, with the discard rate often exceeding 50 percent.
Computing power prices are also rising. Starting April 18, Alibaba Cloud and Baidu AI Cloud both raised prices for AI computing power uniformly. The computing cost for one episode of an AI animated series has climbed from the earlier 17 or 18 yuan (approximately NT$75 or 80) to 80 to 100 yuan (approximately NT$350 to 440). For some higher-quality productions, when you factor in AI tool subscriptions and token computing consumption, production costs can even reach tens of thousands of yuan (approximately tens of thousands of NT dollars). For low-end producers who need to keep per-episode costs down to dozens or a few hundred yuan, directly applying off-the-shelf templates was once the most cost-effective choice—and that is precisely where the homogenization disaster begins.
The Third Mountain: Revenue Shrinks Tenfold — Ten Thousand Plays Are Worth Only One Cup of Bubble Tea
Beyond regulation and costs, what has crushed many bloggers is the cliff-like shrinkage on the revenue side. In the first half of 2025, AI comic drama platforms offered 80 to 100 RMB (approximately NT$350 to 440) per 10,000 plays under revenue sharing, allowing batch production to generate steady profits; by 2026, the same revenue per 10,000 plays had collapsed to 5 to 10 RMB (approximately NT$22 to 44)—a full tenfold decline. Some creators pour in substantial time and computing power, only to end up with a few hundred plays and almost no earnings, leaving the input-output ratio severely imbalanced.
Platforms have also scaled back their traffic support for low-quality bulk content. The algorithms of multiple platforms have shifted to no longer favor frequent updates and batch output, and no longer give AI content exclusive traffic preference. Instead, they put AI comic dramas and live-action short dramas into the same competitive system, keeping only content with high completion, high engagement, and high retention. This has caused about 90% of independent content to completely lose exposure, where free updates and continuous output can no longer earn organic traffic.
Platform cracks down: 2.6 million pieces of low-quality content and over 20,000 accounts removed.
The wave of stopped updates is just the surface; the harsher reality is the platform’s behind-the-scenes purge. According to public platform data, over the past year, more than 2.6 million pieces of low-quality AI content were removed, and over 20,000 low-quality verified authors were expelled. In the first quarter alone, the Hongguo platform removed a total of 1,718 non-compliant AI comic dramas, and in the subsequent “special crackdown on low-quality comic dramas,” another 3,522 were intercepted and penalized within a week.
The eliminated accounts mainly fall into three types: pure reposters of tool tutorials, matrix accounts that mass-produce “one article in 10 seconds,” and unauthorized derivative-content accounts. The stronger AI tools become, the more users prefer to ask AI directly, diluting the value of intermediary popularizers. Million-follower influencers who grew purely on information asymmetry have lost their foundation.
Hit rate plummets: production doubles, but less than 0.5% break 100 million.
The gap between production output and hit rate best illustrates where this business goes wrong. Last December, there were over 50,000 AI animation dramas streaming, with 96 surpassing 100 million views. By February this year, that number had surged to 120,000, yet fewer than 150 works broke 100 million views. Production more than doubled, but the hit rate was cut in half. In the first half of 2026, new AI short dramas across the entire internet reached 221,900 — an average of over 1,200 new releases per day — yet only 1,055 surpassed 100 million views, putting the break-out rate at a dismal 0.47%.
When ten thousand “Reborn as XXXX” stories flood the feed at once, the ending is already written. Late last year, “physiological aversion to AI faces” trended on Weibo with nearly 4 million views; blended faces, stiff expressions, cookie-cutter character settings, and homogeneous plots have quickly exhausted viewers’ patience.
The bubble has burst, but the top players are still making money.
If you only focus on the wave of halted updates and the chaos, it’s easy to think AI comic series have “gone cold.” The other side of the story is that top players are still quietly making money. A post-95s couple from Bazhong, Sichuan, who created *Exiled to the Border, the Abandoned Wife Clears Land and Raises a War God*, have seen 2.127 billion monthly views since it launched in June, hitting 4.1 billion views in two months. They’re in no rush—they fix the visual glitches AI generates frame by frame, license proper IPs from the Hongguo copyright library, and spend their own money on professional equipment and premium AI tools. Top-tier projects like *The Legend of Ten Thousand Demons Season 8* and *Super Pampered at Four and a Half* still pull in tens of millions in popularity points (in short, if you put real effort into the content rather than just dreaming of brainlessly churning out junk with AI, it still pays off).
The industry is clearing out at a staggering pace: in the first quarter of this year, the number of active production houses in China’s AI comic drama sector dropped from 1,216 to 698, with over 40% of teams exiting the market. But the vast majority of those who left were speculators churning out low-quality批量 content to ride traffic waves. Luan Tianxing, CEO of Xin Vision Media, said in an interview that the declining hit rate of AI short dramas is the inevitable result of multiple overlapping factors—a “necessary growing pain on the path from crude expansion to high-quality development,” a retreat of the bubble and a return to value, not a failure of the product itself.
Conclusion: AI is just a tool; the days of easy money are over.
This wave of pauses and cancellations has pushed the AI content industry to a watershed moment. The more widespread the tools become, the more valuable human judgment, experience, and aesthetic sense are; those who treat AI as a producer and just want to churn out content at scale for arbitrage are doomed to be abandoned by both algorithms and the market. The winners still standing rely on original depth, human warmth, and the ability to integrate AI into real work environments. As with every past wave of technology—from euphoric influx, to bubble burst, to rational return—AI-generated content is passing through the turning point of this cycle. A bubble bursting isn’t necessarily a bad thing; this industry has finally started to take itself seriously.
Source: KOCPC Chinese