On April 20, 2026, at the iQiyi World Conference, the names of over 100 artists, including Chen Zheyuan, Cheng Lei, and Zeng Shunxi, were tagged as part of an AI Artist Library, marking the launch of a bold and controversial industry experiment. However, within hours, artists such as Zhang Ruoyun, Yu Hewei, and Wang Churan issued statements through their studios denying any AI authorization agreements, causing public opinion to instantly reverse. Hashtags like "iQiyi must have principles even if desperate" trended on Weibo. In an emergency clarification released late at night, iQiyi stated that inclusion in the library only indicated willingness to cooperate, with specific projects requiring separate authorization negotiations, following the same process as collaborations with real actors. Overnight, a high-profile technological innovation announcement evolved into a nationwide debate concerning industry ethics, artist rights, and platform strategy.
This debate reflects far more than the handling of a public relations crisis. It acts as a prism, revealing from different angles the survival anxiety of long-form video platforms in an era of market saturation, capital's inherent drive for cost reduction through technology, and the enduring tension between artistic creation and technological efficiency.
The Cost-Reduction Dilemma and Data Challenges: Dual Drivers of the AI Artist Library To understand why iQiyi is so urgently promoting its AI Artist Library, one must first recognize the dual pressures of finance and competition it faces.
Financially, the尴尬 of increasing revenue without increasing profits continues to intensify. For the full year 2025, iQiyi's total revenue was 27.29 billion yuan, a decrease of 7% year-over-year, marking the second consecutive year of negative growth after a 7.97% drop in 2024. More concerning for the capital market was the cliff-like decline in profitability—Non-GAAP operating profit was only 640 million yuan, shrinking approximately 73% from 2.36 billion yuan in 2024. Under GAAP standards, net income attributable to common shareholders turned from a profit of 764 million yuan in 2024 to a loss of 206 million yuan. The narrative of "four consecutive years of operating profitability" proudly cited by iQiyi CEO Gong Yu is gradually losing focus in the face of collapsing profits.
Core businesses showed weakness across the board: membership service revenue was 16.81 billion yuan, down about 5% year-over-year; online advertising revenue was 5.19 billion yuan, down about 9%; and content distribution revenue was 2.5 billion yuan, down about 12%. On the US stock market, iQiyi's share price has shrunk by over 97% from its peak, with a market capitalization loss exceeding $30 billion. The secret application for a Hong Kong listing submitted in March 2026 was both a passive move to seek liquidity and an active attempt to rewrite its story.
On the competition front, short-form video and micro-drama platforms continue to divert user time and attention. While platforms like Red Date Short Drama gain ground with low-cost, high-frequency content production models, the high-investment, long-cycle model of long-form video platforms appears increasingly cumbersome. More critically, the appeal of the content itself is problematic—whether for series or variety shows, the number of hit titles has significantly decreased, and user willingness to pay remains weak. As one commentary pointedly noted: if the core problem is that the content is "not good to watch," then even if costs are reduced to one-tenth, the resulting product will still be unappealing, and audiences will not pay for it.
Technologically, AI offers unprecedented possibilities for cost reduction. In Gong Yu's proposed "Media 112 Rule," AI would reduce unit content costs by one order of magnitude, increase the number of creators by one order of magnitude, and achieve a two-order-of-magnitude growth in the number of works. Under the traditional film and television model, an actor can typically participate in about 2 to 4 works per year; an AI digital replica could increase annual output to 14 works. Specifically regarding cost structure, AI technology could reduce film and television production costs by 30% to 50% and shorten production cycles by over 40%.
From an industrial logic perspective, the launch of the AI Artist Library is not an isolated event but a key piece in iQiyi's systematic AI strategy. In 2025, iQiyi applied for 880 invention patents, more than half of which were AI-related. It had previously launched "Script Workshop"—capable of condensing millions of words of a novel within 30 minutes; built an Agent-based "AI Dubbing Factory" that translated 20,000 episodes of long series in a single quarter; amassed over 30,000 digital assets in its self-built digital asset library; and saw a 125% year-over-year increase in virtual production projects. The AI Artist Library is the crucial puzzle piece connecting "people" and "content" within this full-chain AI integration—digitizing the actor, the core production factor, and embedding it into the intelligent production闭环 from script to finished product.
However, the debut of the AI Artist Library also exposed deep-seated challenges in data assetization. High-precision collection of artist likenesses, voices, and performance data, along with model training, are technical prerequisites for deploying digital replicas. Current AI artists still exhibit an "eerie artificiality" in their expressions, movements, and line delivery. From model creation to content implementation, the entire process relies heavily on powerful computing support. More棘手ly, data collection requires the active cooperation of artists for facial scans and motion capture, which is precisely the aspect artists are most cautious about—once high-precision digital assets are handed over, the core concern for all artists and their agencies is whether the platform could bypass them in the future for unlimited reuse.
A deeper矛盾 lies in how the launch of the AI Artist Library also revealed structural deficiencies in iQiyi's content data. As a long-form video platform, its core data asset is the film and television content itself, not user creation behavior or social relationships. This inherent weakness risks relegating it to a "material supplier" role in the AIGC era—where technology companies control the models, and traffic platforms control distribution. While artist digital replicas might boost content output and reduce production costs in the short term, without fostering genuine creator community activity and user social stickiness, the vision of "decentralization" could merely become another form of centralized resource pooling.
Industry Restructuring and Ethical Controversies: Testing the Boundaries of Technological Idealism When iQiyi announced its AI Artist Library plan, its strategic narrative extended far beyond the four words "cost reduction and efficiency enhancement." Gong Yu painted a grand picture of industry restructuring: AI will directly change the content itself, thereby pushing the platform to transform from a centralized video website into a decentralized social media platform characterized by "creator and user communities + head content production."
Within this narrative, iQiyi, through its Nadou Pro platform, fully opens its nearly 20-year accumulated IP library, artist library, digital asset library, and commercial resources to creators. Creators can complete the entire production process from script generation to finished output on the platform, with works directly entering the distribution system and participating in revenue sharing. The platform's role shifts from "content purchaser and distributor" to "provider of ecological infrastructure"—this is both an active choice in response to the explosive growth of content supply in the AIGC era and an inevitable path for long-form video platforms to break free from the cycle of "content investment - attract new subscribers - reinvest in content."
From the perspective of industry competition, iQiyi's AI transformation is not an isolated case. Tencent Video focuses on vertical short dramas and AI-driven technology, with Tencent Vice President Sun Zhonghuai clearly identifying AI as one of two key strategic variables. Youku's parent, Orca Entertainment Group, is also actively embracing AI, accelerating the integration of an "online + offline" full-scenario entertainment ecosystem. The three major platforms' convergent choice of AI as a breakthrough direction reflects the collective anxiety of the entire long-form video industry facing user growth ceilings. The difference is that iQiyi has chosen the most radical path—elevating AI from an "assistant tool" to a "core productive force," and transforming artists from "collaborators" into "data assets."
However, the collective artist denial incident revealed a fundamental obstacle to implementing this strategy. When artists like Zhang Ruoyun, Yu Hewei, Li Yitong, and Wang Churan publicly denied AI authorization, it exposed not only a communication failure but also deep-seated disagreements on core issues such as data rights ownership, profit distribution mechanisms, and legal safeguard frameworks.
Regarding data rights, once an artist's likeness, voice, and performance data are collected, they become high-precision digital assets owned by the platform. The boundaries of ownership and usage rights for this "digitized performance capability" are extremely模糊. Although iQiyi explicitly adopts a "project-based authorization model," where artists only grant authorization for specific series roles and retain character veto rights, once the data is in the platform's hands, could it, in the long run, create a bargaining advantage over the artists? Are actors "digging their own graves"—cultivating potential competitors that might replace them?
Regarding profit distribution, how to price the performances of AI digital replicas, determine revenue-sharing ratios, and set authorization periods currently lacks industry standards and precedents. Although iQiyi states that the "rules are consistent with the collaboration process for real-actor film and television projects," AI performance and real-actor performance are fundamentally different—the former involves the reuse of data assets, while the latter involves the input of creative behavior; their cost structures and value creation cannot be simply compared.
Legally, AI artists involve multiple rights such as portrait rights, voice rights, and performer rights. Article 1018 of the Civil Code stipulates that the making, use, or public disclosure of a portrait right holder's portrait requires their consent. Article 1023 explicitly states that the voice rights of natural persons are protected by law. A prior judicial case ruled that AI "cloning" of a celebrity's voice for livestream sales constituted infringement, resulting in a 120,000 yuan compensation. During this exploratory period with no established industry standards, the relationships of rights and responsibilities among platforms, artists, and creators are fraught with uncertainty, and negligence at any stage could trigger legal disputes.
From the audience's perspective, the questions are equally尖锐. When all platforms adopt similar facial synthesis and emotion generation models, the expressions, movements, and line delivery of AI artists will inevitably become increasingly homogenized—whether in period dramas or modern series, AI-portrayed sadness might uniformly involve frowning, and anger could easily default to identical wide-eyed stares. This reduced cost cannot drive user growth and might instead accelerate audience aesthetic fatigue. If users believe AI artists impair their viewing experience, can they claim rights from the platform? Under the current consumer protection framework, there is no clear answer to this question.
A deeper question is: What is the essence of film and television creation? Chen Daoming's improvised "scolding the ministers" monologue in "Kangxi Dynasty" became a highlight of the series; the subtle shift from despair to determination in Xu Zheng's eyes in "Dying to Survive" stemmed from the actor's deep empathy with the character's situation. These inspired moments did not come from scripted directions but resulted from actors "precipitating" their life experiences into their performances. AI can precisely replicate actions and expressions but cannot "feel" the role or provide improvisational reactions. Gong Yu himself admitted: "At the current stage, good works still require real actors." He also raised a more intriguing question—without the enrichment of technology, would works that are completely 100% real be designated as World Cultural Heritage, becoming "intangible cultural heritage" years later?
iQiyi's AI Artist Library is essentially a radical experiment driven by both financial pressure and technological opportunity. From a business logic standpoint, using technology to compress costs and increase capacity is understandable—especially when the platform has seen revenue decline for two consecutive years and profits plummet by 70%. Any attempt to generate增量 deserves consideration.
However, from an industrial logic perspective, the boundary of technology lies precisely in its inability to answer the fundamental question: "Why do people create, and why do they watch?" The quantity of works might increase by two orders of magnitude, but what truly remains in the audience's hearts are always those moments imbued with "human touch." iQiyi's long-form video困境 stems from the hollowing out of content appeal. Placing all hope solely on AI is tantamount to using technological means to掩盖 a deep-seated crisis in content strength.
In the long run, the success or failure of the AI Artist Library will not depend on how advanced the technology is, but on whether it can find a balance between efficiency and creativity, data assets and human dignity. If platforms, creators, and artists can build an ecosystem of mutual trust and shared benefits, AI could well become a powerful supplement to the film and television industry. However, if the pursuit of cost reduction overlooks humanistic values, the final output might be vast quantities of "algorithmic content" rather than truly moving artistic works. iQiyi's AI gamble is not just a bet on a technological route, but an answer to the fundamental question: "What exactly is film and television creation?"
Comments