On April 21, iQIYI clarified its Nadou Pro AI actor library, stating it is a standardized collaboration platform for AIGC creators and actors to communicate efficiently, and that no actors were added without consent. The announcement followed controversy sparked when iQIYI Senior Vice President Liu Wenfeng revealed that over 100 actors had signed on to the platform during the iQIYI World Conference.
The disclosure immediately triggered public concern about AI replacing human performers. On the same day, multiple actors—including Yu Hewei, Zhang Ruoyun, and Wang Chuzhen—issued statements denying AI authorization after being mentioned in social media posts about the platform. Upon confirmation, these three actors were found not to be part of Nadou Pro’s actor library; their denials addressed circulating rumors.
What Is Nadou Pro’s Actor Library?
According to iQIYI’s statement to Peng Pai News, the Nadou Pro actor library allows AIGC creators to select actors from the database and communicate quickly with them regarding collaboration rights and execution details. The platform, which entered trial commercialization in March 2025, is a professional-grade film and television production platform integrating large language models, intelligent agents, iQIYI IP assets, digital resource libraries, commercial partnerships, and creator communities.
Confirmed participants listed by Liu Wenfeng include Ma Su, Cheng Taisheng, Chen Zheyuan, Zeng Shunxi, stand-up comedian Fang Zhuren, and comedian Jiang Long. Only Chen Zheyuan’s team has denied signing AI-related authorization.
Why iQIYI Is Pursuing AI Actors
iQIYI CEO Gong Yu stated that AI is significantly reducing film and television production costs, shortening production cycles, and lowering production difficulty. These improvements, he argued, will accelerate growth in creator numbers, work output, and user consumption, driving explosive industry expansion.
Based on this assessment, iQIYI is undergoing decentralization transformation to become a non-centralized social media platform combining “creator and user communities with premium content production.” Nadou Pro serves this decentralized creator ecosystem.
In public remarks, Gong Yu noted that actors work “four to five months with 13-14 hour daily shifts and no personal life.” He suggested AI could enable performers to work like “ordinary office workers with some personal time, accepting 4 projects annually instead of 2.” Gong Yu also posed a philosophical question: whether purely authentic works without technological enrichment might eventually be designated as world cultural heritage or intangible cultural relics.
These comments generated public backlash, with social media users responding: “If actors become AI, what warmth remains in artistic works?” and “AI was created to serve humanity, not replace it.”
iQIYI clarified to Peng Pai News that entry into the Nadou Pro actor library indicates an actor’s willingness to engage with AI film projects, but participation in specific projects or roles requires separate negotiation and authorization—mirroring traditional film collaboration processes. Gong Yu emphasized during the conference that rules mirror live-action production: actor consent, managed through talent agents and agencies, is required per project and role. Authorization for one project does not imply consent for others.
Legal Framework: AI Likeness Rights and Infringement Standards
As AIGC technology barriers lower, multiple actors have filed complaints about unauthorized use of their likenesses, voices, and images in AI-generated content, particularly in short drama productions. In April alone, actors Yixuan Qianxi, Tan Jianci, Gong Jun, Deng Wei, and Zhang Jingyi issued studio statements condemning AI infringement and demanding removal of unauthorized AI content.
Beijing Jiawei Law Office partner Zhao Zhanling clarified to Peng Pai News: “According to the Civil Code, determining likeness right infringement depends on whether the AI-generated likeness is identifiable as a specific person and whether legal authorization was obtained. Commercial use is no longer a necessary condition. For voice, likeness protection rules apply by analogy: if AI-synthesized voice allows the general public or relevant professionals to associate it with a specific person based on timbre, intonation, and pronunciation style, it falls within protection.”
Beijing Jingshi Law Office attorney Meng Bo added that likeness must be externally identifiable with recognizability as a prerequisite. “Determining AI infringement requires assessing whether the actor implemented specific conduct and possessed corresponding subjective intent, beyond the likeness’s recognizability.”
Regarding “face-blending” (merging multiple actors’ facial features) as potential infringement avoidance, Zhao Zhanling explicitly rejected this defense: “If the AI-generated face retains core features of a specific actor, or accompanying text or environmental context identifies that actor, infringement still occurs.”
Zhao Zhanling also highlighted systemic industry problems: “Current AI actor sectors exhibit vague authorization and unclear boundaries. Traditional performance involves one-time labor compensation; AI authorization transfers personality element commercialization rights. Once authorized, it may long-term impact an actor’s market value. Risks including model fine-tuning, data leaks, and unauthorized secondary training remain difficult to control under existing frameworks.”
On user rights regarding poor AI performance, Zhao Zhanling assessed legal grounds as weak: “Unless the platform explicitly promises ‘AI performance meets human standards’ and severely fails to deliver, audiences struggle to claim breach of contract or damages based solely on poor AI performance quality.”
Industry Response and Regulatory Direction
On April 2, the China Broadcasting and Television Association’s Actor Committee issued a statement condemning current AI face-swapping, voice cloning, and unauthorized film material alteration. The statement demanded that short video, livestream, and film distribution platforms enforce content moderation responsibility, establish long-term AI performance content authorization verification mechanisms, and immediately audit and remove existing infringing works. AI technology developers and service platforms must also strengthen pre-publication review obligations, strictly verifying actor likeness, voice, and image materials for authorization credentials.
Zhao Zhanling proposed that future industry standards should establish a “authorization consent plus mandatory labeling” dual safeguard, using “recognizability” to define infringement boundaries while allowing technical innovation space through detailed authorization and platform compliance obligations.
Business Performance Context
On February 26, iQIYI reported 2025 fiscal Q4 results with revenue of 6.79 billion yuan, up 3% year-over-year. Net loss attributable to iQIYI was 5.8 million yuan, compared to 190 million yuan loss in the prior-year period. Non-GAAP operating profit reached 140 million yuan. Membership service revenue was 4.11 billion yuan (flat year-over-year), online advertising revenue was 1.35 billion yuan (down 6%), content distribution revenue was 788 million yuan (up 94%), and other revenue was 550 million yuan (down 18%).
