17 September 2026 News

Entertainment : In the film, television, and short-form drama sectors, digital actors formally participating in live-action series has become a new trend. Domestically, cases have emerged where digital human images are highly aligned with plot narratives and formally participate in live-action series. Cyberspace Administration of China has accordingly drafted the “Digital Virtual Human Information Service Management Measures (Draft for Comment)” to regulate related businesses. Facial authorization transactions have subsequently emerged: Lianyungang Haizhouqu (连云港海州区) launched the nation’s first AI facial and vocal authorization traceability verification platform “Yuanlian Zhazha (元脸查查)”. Yuan, a woman from Sichuan working in Shanghai, authorized her AI avatar to perform as the female lead in a Douyin short drama titled “Piao Piao,” cumulatively earning 1,000 yuan in compensation. Zhong, a film and television industry professional from Lianyungang, also used the platform to have his AI image featured in two episodes of a short drama. An investigation by Beijing News also discovered facial trading platforms including “Xinzhao (新爪),” “Yuanxiang Xinsheng (元相新生),” and “Huimeng · Xingxiang (绘梦·星相),” with listing prices ranging from 50 yuan to tens of thousands of yuan. Some platforms have agreement terms skewed in favor of the platform, and cases have even appeared where minors aged 16 or 17 can authorize their likenesses, raising legal risk concerns. Non-celebrity authorization remains equally active, with facial authorization fees for an AI short drama averaging around 100 yuan. At the celebrity level, Qi Wei launched AI digital avatar “Han Qingxia,” releasing an apocalyptic-themed AI animated short drama “Final Midsummer” and introducing the project at the Venice Film Festival’s AI Image International Round Table, claiming that proactive authorization aims to curb piracy. However, controversy arose after the promotional video was accused of “overstepping boundaries” and “monetizing faces.” Against the backdrop of the formal implementation of the “Micro Short-Form Drama Development Management Measures,” celebrities including Wang Zuxian, Qi Wei, Wu Qihua, and You Benxiang have authorized others to use their likenesses to create “digital avatars” and enter the AI short-drama market, while grassroots extras have experienced income drops exceeding 70 percent across the board. Shenzhenshi Laonian Xiehui (深圳市老年协会) partnered with local technology teams to create a silver-haired themed AI live-action fusion short drama “Old Partners,” where elderly people without acting experience participate through digital avatars, telling the story of retired university students reforming a band to pursue their dreams. Gaming streamer “Yitiao Xiaotuan” resumed broadcasting after a two-year hiatus, launched a guild, and produced AI short dramas. The streamer also partnered with Ji Meng AI (即梦AI) to launch the “Xiaotuan Dream Secondary Creation Competition,” opening personal digital avatar images for secondary creation by users across the internet, with total prize money exceeding 500,000 yuan. Huanrui Century introduced conversational AI digital humans in the series “Lotus Tower,” with technology support provided by N Shijie (N世界), exploring a promotional approach combining IP, digital humans, and AI technology. Quzhou (衢州) used AI short dramas and AI digital humans in united front cultural monthly activities to recreate stories of shared purpose along the red trade routes of western Zhejiang.

Cultural Tourism : The cultural tourism industry is introducing digital humans and virtual tour guides on a large scale. Hangzhou Gongshuqu (拱墅区) leverages Guojia Rengongzhineng Yingyong Zhongshi Jidi (国家人工智能应用中试基地) (cultural tourism) to build a complete ecosystem covering computing power, data, models, applications, and scenarios; Shenzhen Gankeng Guzhen (甘坑古镇) launched an “AI Digital Human Li Bai Interactive Themed Show” enabling improvisational Q&A; Nanchang Tengwangge (南昌滕王阁) “AI Tour Guide Wang Bo” can provide explanations based on visitor locations and transform into an “examiner” to score visitors reciting the “Preface to the Pavilion of Prince Teng”; Hengyang Chuanshan Shuyuan (衡阳船山书院) an AI digital human project allows visitors to converse with a virtual “Mr. Chuanshan” and trigger AR annotations through gestures. Suzhou Wenbohui (苏州文博会) On iFLYTEK, an “AI Poetry and Painting Jiangnan Garden Party” lets digital human Li Bai compose poetry improvisationally, while robot “Qiuxiang” performs the Kunqu opera “The Peony Pavilion” and supports multilingual communication. In Fujian, Yangguang Xueyuan (阳光学院) a student team designed the Shipping and Politics Museum’s explanatory AI image as a Q-version digital human “Chuan Xiaozhi” of Zuo Zongtang, and after upgrading to version 2.0 can provide real-time navigation and answer questions about shipping history. The 23rd China–ASEAN Expo for the first time set up an “AI Marketplace,” live demonstrating multilingual digital humans bringing goods to market; the smart bus station display in the South A Community Zone can interact with digital humans through voice to inquire about transportation information. Xian Jinghe Xincheng (西安泾河新城) AI Digital Youth City received top social media influencers from Laos, using high-precision facial modeling and AI intelligent modeling technology to generate exclusive digital human images for the influencers navigating through Tang Dynasty costume scenes; Laotian influencer Seng Aphay Vannaphone remarked this was the first time she “appeared in a movie” this way. Shandong Taian Taishanqu (泰安泰山区) Baima Shi OPC Shequ (白马石OPC社区) positions itself in the AI cultural tourism track, activating Mount Tai culture through AI short dramas and digital human applications as entry points. Kunshan Xiqu Baixi Bowuguan (昆山戏曲百戏博物馆) In its XR exhibition “Garden Fantasy Play,” virtual recommendation hosts “Xiaoxiao Sheng” and “Xiaohua Dan” assume research study tour guide functions; additionally, the airport expressway transportation exhibition hall has introduced AI digital human guided tours providing interactive explanation services.

Marketing : The compliance boundaries of AI virtual humans in advertising and marketing have become a hot topic. The AI virtual human “Fang Taozi” promoted a certain brand’s colored contact lenses in a consumer voice, claiming “wearing them all day is very comfortable.” Because virtual humans lack the objective conditions to genuinely test products and colored contact lenses are classified as Class III high-risk medical devices, the advertisement was eventually taken down and triggered regulatory intervention. Commentators pointed out that AI virtual humans do not have legal personhood and cannot serve as endorsers to recommend products. In 2026, amid intense competition from celebrity endorsements in the beauty industry, Hua Xizi (花西子) launched an exclusive AI digital endorser, Baiqueling (百雀羚) launched the AI digital model “Bailing,” and the Chinese-style virtual human Ling once used “personal experience” to promote lipstick, triggering widespread questioning about whether virtual humans can authentically experience products. Polaiya (珀莱雅) and other brands, even with matrices of both real and virtual endorsers, have not seen significant performance improvements. Caixun Technology Co., Ltd. disclosed that its 3D hyper-realistic intelligent video generation platform and “AI Zhiwu (AI智舞)” have been deployed in multiple scenarios, previously created virtual digital human system imagery and operation tools for China Mobile, built intelligent digital human services for Suyanyun (苏研云)’s capability center, constructed an AI digital human platform for Zhongyihu (中移互), applied to digital human livestreaming, video customer service, and smart cultural tourism marketing. Following its signing of Su Bingtian as its first real-person endorser in 2021, Shunfeng Tongcheng (顺丰同城) has formally launched the virtual IP endorser “Tongxiaoxia,” positioned as a service-oriented digital persona with five identities: action-taker, considerate partner, merchant partner, knight captain, and workplace companion. Additionally, Lanse Guangbiao (蓝色光标)’s subsidiary Blue Universe Digital Technology metaverse marketing platform reuses the group’s BlueAI GEO (Generative Engine Optimization) capabilities to provide AI-structured content optimization services for scenarios including virtual human IPs, digital collectibles, and virtual exhibitions.

Enterprise : On the capital market, multiple A-share listed companies have disclosed product progress around the “virtual digital human” concept. Tianyou Dejiu (天佑德酒) launched the digital virtual character “Li Ling’er,” integrating regional characteristics and the history and culture of highland barley liquor into the digital human image; Xin Guomai Shuzi Wenhua (新国脉数字文化) introduced enterprise digital-smart virtual humans “Irene” and “Xiao Ling” based on prototypes of the general manager and employee representatives, used for brand and business promotion; Beijing Guoji Renliziben Jituan (北京国际人力资本集团)’s subsidiary Waiqi Shuke (外企数科) leverages big data and AI algorithms to launch digital human images such as “HR Miss,” “Aili,” and “Zheng Duoduo” on China International Fair for Trade in Services (CIFTIS) that enable real-time dialogue and possess physical feedback capabilities; Shenzhenshi Tianwei Shixun (深圳市天威视讯) integrated AI large model technology to launch TV AI voice digital humans and intelligent customer service; Hebei Guangdian Wuxian Chuanmei (河北广电无线传媒) developed an AI-based digital human interaction system, completing image capture and modeling, skeleton motion rigging, and interactive scenario development, seamlessly integrable into IPTV set-top boxes, OTT, and other pan-screen applications; Mango Excellent Media disclosed that its technology division has adapted to domestic computing power systems, supporting multi-channel real-time video review, AI editing, and virtual human rendering for live variety shows such as “Singer 2026” and “Riding the Wind 2026”; Wondershare laid out innovative businesses including virtual digital humans, virtual scenarios, and virtual live streaming years ago. Kuaiyu Keji (快语科技)’s publicly disclosed AI digital human virtual human real-time interaction solution shows it unveiled five major modules—image modeling, voice generation, animation rendering, audio-visual synthesis, and dialogue interaction—applied to scenarios such as cultural tourism guidance and recitation assessment, though specific engineering metrics such as rendering frame rate and latency still require field testing and confirmation in deployment environments. In the live-streaming e-commerce sector, JD.com’s digital human “Caixiao Donge” attracted over 13 million views at its debut in 2024, with broadcast volume increasing tenfold year-over-year by the first quarter of 2026; Bilibili’s “AI Angel” and “Mu Ji Meng” interact with audiences as AI virtual hosts, yet irregularities have also occurred where individuals used AI synthesis to create fake livestreams of “Zhang Wenhong” selling protein bars, moving over 1,200 units. Jinan (济南) a man spent 17,800 yuan purchasing AI digital human live-streaming services from an information company, expecting to replace himself in live shopping livestreams, but because the AI virtual human’s on-air product explanations were repeatedly judged as violations and banned by the Douyin platform, ultimately experienced a district people’s court judgment ordering the other party to return 9,000 yuan; the judge reminded consumers to be vigilant against AI livestreaming project promotions claiming “low investment, high returns, passive income with guaranteed minimum.”

Public Sector : In government affairs and public service scenarios, digital humans are being deployed for safety promotion and customer service. Chengdu (成都) The opening ceremony of the 2026 National Cybersecurity Promotion Week series of activities featured a virtual human delivering opening remarks, with interactive experiences including AI face-swapping and deepfake recognition; Wuhan (武汉) in the cybersecurity promotion week’s metro-themed campaign, the digital human “An’an” appeared as the protagonist in science education animations playing on loop on metro televisions and platform screens, explaining high-frequency risks such as AI face-swapping fraud and fake customer service refund scams. Zhongguo Guohang (中国国航) A “customer service holographic virtual digital human” and other digital-intelligence service products were showcased at the Fourth Zhongguo Hangxie (中国航协) Aviation Conference. At Jiangsu Province’s “Artificial Intelligence + Medical and Health” series of themed salons, Xuzhou Yikedaxue (徐州医科大学) introduced its practice of promoting data flow to grassroots levels, building “digital humans going to rural areas,” and digital tiered diagnosis and treatment models, bringing smart healthcare to county and village areas.

Technology : Shengshu Technology released a new generation real-time video generation model Vidu S2, led by Zhang Jindao, a doctoral student of Professor Zhu Jun from Tsinghua University and head of Shengshu Technology streaming video generation and inference, with end-to-end research and development completed just 69 days after the previous generation Vidu S1 was released. Vidu S2 - Avatar (Vidu S2-Avatar) supports 720P real-time digital human interaction output at 25 to 42 FPS, can execute large-scale motions such as dancing, and allows users to submit reference images in real time during interaction to change clothing, props, or scenes; Vidu S2 - Editing (Vidu S2-Editing) can perform real-time editing of live-action video streams including outfit changes, character swaps, background changes, and style transformations while maintaining continuity of original movements and camera trajectories; additionally, the team has explored dynamic reference image control based on VLM Agent, as well as extended real-time generation to spatial video capabilities for VR headsets. Technical reports show that the model achieved industry-leading scores in evaluations such as StreamAV-Bench and Sparkle-Bench. Upon release, it was made fully available for experience and API access, marking what is considered a new phase of AI video transitioning from “offline generation” to “real-time interaction.”

Ethical Contexts : Legal and ethical discussions surrounding AI virtual humans and digital cloning technology are intensifying. Cyberspace Administration of China Four departments jointly issued the “Measures for Labeling AI-Generated and Synthesized Content,” which requires that digital human videos display prominent disclaimer labels at the beginning, prohibits deletion or concealment of watermarks, and mandates explicit authorization from individuals whose photographs or voices are used to clone digital humans, with commercial use requiring written agreements. Zuigaorenminfayuan (最高人民法院) The Supreme People’s Court released typical cases on internet rule of law, including the case of Li Moujin v. a cultural company for network infringement liability disputes, in which an e-commerce livestreaming host used video clips of Li Moujin’s speech without authorization and paired them with AI-synthesized audio highly similar to Li’s voice patterns to conduct “impersonated celebrity product promotion.” Beijing Hulianwang Fayuan (北京互联网法院) The first-instance court ruled that the cultural company must issue a public apology and pay 120,000 yuan in compensation, and the Supreme Court clarified that e-commerce operators have an obligation to review promotional content and bear joint liability with livestreaming hosts when constituting joint infringement. Zuigaorenminfayuan (最高人民法院) The concurrently released “Opinion on Lawfully Adjudicating Cases Involving Artificial Intelligence Disputes” further clarifies that unauthorized use of AI-generated virtual images of others for public commercial purposes constitutes infringement of portrait rights, and unauthorized replication of others’ voices for commercial scenarios constitutes infringement of voice rights, while introducing a new mechanism for personality right infringement injunctions to enable rapid removal of infringing content by victims. A 65-year-old woman utilized AI face-swapping and scene synthesis technology to fabricate a false persona of a “post-1990s doctor at a top-tier hospital,” maintained romantic online relationships with more than ten men including Liu Hui, and defrauded Liu Hui of 170,000 yuan over five months, exposing new risks of AI technology lowering the threshold for online fraud and batch-generating high-quality fake personas.