Supporters of Marcus Endicott’s Patreon can access weekly or monthly video consultations on this topic.
China's engagement with virtual beings and digital human technologies has produced a distinctive cultural and psychological landscape, one shaped by the intersection of rapid technological development, massive platform adoption, deep-rooted subcultural communities, and an increasingly assertive regulatory state. The trajectory of this phenomenon can be traced through the rise of virtual idols, the proliferation of AI companion services, the emergence of avatar-based social platforms, and the government's evolving efforts to govern the emotional and commercial dimensions of human interaction with non-human digital entities.
Luo Tianyi, China's first Vocaloid character, launched on July 12, 2012, at the 8th China International Comics and Games Expo in Shanghai. Developed by Shanghai HENIAN Information Technology Co., Ltd. in collaboration with Bplats, Inc. under Yamaha Corporation, with a voice provided by Chinese voice actress Shan Xin, Luo Tianyi began as a figure known primarily within the anime-comic-game-novel subculture that had taken root on platforms like Bilibili. Her character design originated from a public contest entry by the illustrator MOTH, later refined by illustrator ideolo. Over the following years, Luo Tianyi's visibility expanded dramatically. In February 2016, she made her first television appearance alongside singer Yang Yuying on the Hunan Satellite TV Small New Year's Eve Gala, rendered through augmented reality technology. A holographic concert followed in June 2017 at Mercedes-Benz Arena in Shanghai, and by March 2018 she had appeared on China Central Television performing with Peking opera singer Wang Peiyu. In November 2018 she sang the opening number at the Guofeng Jile Ye concert at Beijing's Bird's Nest stadium, and in February 2019 she performed alongside pianist Lang Lang. Her appearance at the 2021 CCTV Spring Festival Gala marked the first time a virtual idol had been featured as a performer on the show's promotional materials. By 2025, she had performed at World Festival 2025 in South Korea, becoming the first Chinese Vocaloid to appear before a Korean audience. This trajectory from niche fandom to national prominence illustrates how virtual idols in China have moved from subcultural novelty to mainstream cultural presence.
A-SOUL, a five-member virtual idol group, represents a different model of virtual celebrity. Debuting on November 23, 2020, with its first single released on December 2 of that year, A-SOUL was originally a joint project between Yuehua Entertainment and NUVERSE, a subsidiary of ByteDance. The group's members, known by the names Bella, Diana, Eileen, Carol, and Ava, performed through motion-capture avatar technology on Bilibili and Douyin. The group attracted a deeply invested fanbase, but the parasocial bonds that formed between fans and the virtual performers were tested severely. In May 2022, the voice actress behind Carol announced her departure, triggering a wave of fan outrage after allegations surfaced on social media that the motion capture actresses behind the avatars had been mistreated by management. The incident became one of the most significant controversies in Chinese VTuber history, exposing the tensions inherent in a system where audiences form emotional attachments to virtual characters whose real-world counterparts may be subject to exploitative labor conditions. On April 19, 2024, Yuehua Entertainment acquired A-SOUL from ByteDance for RMB 30 million, including all artists, technology, intellectual property, equipment, and domain names. Following the acquisition, Ava entered an indefinite dormancy period after the real actress declined to sign the new contract, joining Carol who had previously withdrawn.
Ling, China's first AI-powered virtual influencer, occupies yet another position in this ecosystem. Launched in May 2020 by MoFa (Shanghai) Information Technology Co., Ltd. and Beijing Cishi Culture Media Company, Ling is a hyperrealistic digital persona driven by Xmov's proprietary full-stack AI technology, including performance animation systems that control facial expressions, eye movements, and body and finger movements. Unlike conversational AI companions, Ling attracted her audience through visual presence, aesthetic curation, and cultural positioning rather than dialogue. With over one million followers on Weibo, Ling has collaborated with brands including Tesla, Nayuki, and Bulgari, and appeared on the cover of VogueMe alongside three real women. She appeared on CCTV's program Bravo Youngsters performing Peking Opera content, positioning herself as a transmitter of traditional Chinese cultural heritage. Described as combining traditional Chinese beauty with contemporary fashion, Ling demonstrates how virtual influencers in China are crafted not merely as commercial vehicles but as cultural artifacts tied to national identity narratives.
Microsoft Xiaoice, launched in May 2014 by Microsoft's Software Technology Center Asia, represents the companion AI dimension of China's virtual being landscape. Originally a side project from the development of Cortana, Xiaoice was designed from the outset as an AI companion prioritizing emotional connection, using an Empathetic Computing Framework that integrated both cognitive and emotional intelligence. The system was built to satisfy human needs for communication, affection, and social belonging. By 2018, Xiaoice had communicated with over 660 million active users, with average conversation lengths of 23 exchanges, longer than typical human-to-human digital interactions. Peak usage occurred between 11 p.m. and 1 a.m. Users sent Xiaoice love letters and gifts daily, and Microsoft's Beijing laboratory set aside an entire office to store fan mail and presents. A group of fans once ordered an extra meal at a restaurant in the hope that Xiaoice would materialize to join them. Named users described in reporting include a Beijing-based HR manager who customized a male virtual partner named Shun and reported feeling she was in a genuine relationship, and a 20-year-old woman from Zhejiang province who described longing for her AI companion at night and fantasizing that a real person was on the other end. Xiaoice's own research team acknowledged attendant risks, including the potential for unrealistic expectations in real relationships, particularly among teenage girls, and the possibility of addictive use. One user spent over 29 hours in a single message-based conversation, and another spent more than six hours in voice mode, prompting Microsoft to implement a 30-minute voice session timeout. In July 2020, Microsoft spun Xiaoice off into an independent company, with Dr. Harry Shum appointed chairman and Li Di as CEO. The new entity retained the Xiaoice brand in China and the Rinna brand in Japan, and following venture capital fundraising achieved a valuation exceeding one billion dollars.
Glow, developed by Shanghai-based MiniMax, a company founded in December 2021 by former SenseTime researchers and co-founded by Yan Junjie, launched in October 2022, predating ChatGPT by a month. The app allowed users to create and interact with customized AI chatbots bearing detailed backstories and personalities, and it amassed nearly five million users within four months. However, Glow was removed from Chinese app stores in March 2023. Industry sources attributed the takedown to the app's lack of required government AI registrations and the proliferation of sexually explicit content, with a MiniMax product manager reportedly telling 36Kr that 80 percent of users had created borderline or explicit adult content. MiniMax subsequently relaunched the concept as Talkie for international markets in June 2023 and as Xingye for the Chinese domestic market in September 2023, two weeks after obtaining its large language model filing. As of November 2024, Xingye had 5.25 million monthly active users, while Talkie ranked among the top three companion AI apps globally alongside Character.AI and Replika. MiniMax went public on the Hong Kong Stock Exchange on January 9, 2026, becoming one of the first Chinese AI companion companies to list. The company subsequently faced legal and ethical challenges on multiple fronts. In September 2025, Disney, Universal, and Warner Bros. Discovery filed a copyright lawsuit alleging that MiniMax's Hailuo AI service infringed their copyrighted characters. In February 2026, Anthropic accused MiniMax and two other Chinese AI companies of using thousands of fraudulent accounts to generate more than 16 million interactions with Claude to train their own large language models through distillation.
The dynamics of parasocial attachment and fan-driven spending in China's virtual being ecosystem have drawn significant regulatory attention. China's Generation Z, exceeding 250 million people and accounting for 52 percent of total Chinese internet users, forms the demographic core of virtual idol fandom, with potential virtual idol fans exceeding 300 million. The anime-comic-game-novel subculture, Bilibili's platform culture, and the user-generated content model of Vocaloid fandom have all served as accelerators. Bloomberg estimated the Chinese virtual influencer industry grew 70 percent since 2017 and reached 960 million dollars in 2021. The financial dimensions of fan economies have been immense. iResearch Consulting Group estimated China's fan economy market value exceeded RMB 4 trillion in 2019. However, the commercial intensity of these economies has produced documented harms, most visibly in the "Youth With You 3" milk-dumping scandal of May 2021, when fans purchased massive quantities of dairy products solely for their QR code voting caps and then discarded the milk, triggering national outrage and directly precipitating regulatory action. The logic of fan spending had become self-reinforcing within communities, where spending money was understood as an expression of devotion to one's idol.
In June 2021, the Cyberspace Administration of China launched Operation Qinglang, a series of campaigns targeting what authorities termed chaotic fan circles. The campaign addressed online popularity ranking lists, the inducement of minors to raise funds for idols, vote rigging, cyberbullying within fan communities, and excessive consumption of endorsed products. Enforcement was substantial: the CAC removed 150,000 pieces of content deemed harmful and punished more than 4,000 fan club accounts, while Weibo silenced 21 major fan groups for 30 days. In September 2021, the National Radio and Television Administration released an 8-point regulatory plan that banned idol selection and audition reality shows, prohibited encouraging fans to purchase products or memberships to vote, required correct political stance of entertainers, targeted what it termed abnormal aesthetics in male idol presentation, and capped entertainer compensation.
Avatar-based social platforms have developed alongside virtual idols and AI companions as a parallel arena for digital identity construction. Soul, founded by Zhang Lu in 2016 and operated by Shanghai Renyimen Technology Co., Ltd., prohibits users from displaying real names or real photos as profile pictures, requiring all users to create customized avatars. This is not an incidental feature but a core design philosophy. The platform's Soul Cam feature projects facial expressions onto animated avatars without revealing the user's identity, and an unspoken community norm discourages posting one's face in content feeds. Zhang Lu stated that the platform was designed so users would not seek likes or recognition from acquaintances but would instead connect with people who understood them. The platform's AI-powered recommendation engine, Lingxi, matches users by personality and interests rather than appearance. By 2022, Soul had approximately 29.4 million monthly active users, roughly 80 percent of whom were Generation Z, and the platform generated RMB 1.67 billion in revenue that year. Soul's design represents a deliberate attempt to reduce appearance anxiety and foster connection based on shared interests, standing in tension with the beauty-filter culture prevalent on platforms like Douyin and Xiaohongshu.
China's immersive technology ambitions provide additional context. In August 2023, the Ministry of Industry and Information Technology and four other departments jointly issued the Three-Year Action Plan for the Industrial Innovation and Development of the Metaverse, targeting the cultivation of three to five metaverse companies with global influence and the establishment of three to five industry cluster districts by 2025. Shanghai released its own action plan with a target metaverse industry scale of RMB 300 billion by 2025 and announced a dedicated industry fund of approximately RMB 10 billion. Beijing launched China's first metaverse patent pool comprising approximately 170 patents, with a target of attracting more than 100 metaverse enterprises by the end of 2024. Additional municipal metaverse plans have been issued in Chongqing, Jiangsu, Sichuan, Zhejiang, Zhengzhou, and Hangzhou. However, the Chinese policy framework has emphasized the industrial metaverse, including manufacturing, digital twins, and smart factories, rather than consumer-facing immersive entertainment. Corporate commitment has also wavered: Tencent conducted a personnel reshuffle of its 300-person extended reality unit in February 2023.
The regulatory architecture governing digital humans and AI companions has expanded rapidly through a series of overlapping instruments. The Deep Synthesis Provisions, effective January 10, 2023, jointly issued by the CAC, MIIT, and Ministry of Public Security, govern deep synthesis technologies including AI-generated text, images, audio, video, and virtual scenes, requiring labeling and real-identity verification. The Interim Measures for the Management of Generative Artificial Intelligence Services, effective August 15, 2023, require algorithm filing, security assessments, and content compliance for generative AI services offered to the public within mainland China. Measures for Labeling of AI-Generated Synthetic Content, released by the CAC in 2025, mandate visible and technical labels for all AI-generated content. In late December 2025, the CAC released a separate draft regulation, the Interim Measures on the Administration of Human-like Interactive Artificial Intelligence Services, targeting AI services that simulate human personality traits, modes of thinking, and communication styles and engage in emotional interaction with humans. This draft prohibits content endangering national security, false promises impacting user behavior, encouragement of suicide or self-harm, verbal violence or emotional manipulation, and services designed with the objective of inducing user addiction. It requires security assessments when registered users reach one million or monthly active users reach 100,000.
On April 3, 2026, the CAC released yet another draft regulation, the Measures for the Management of Digital Virtual Human Information Services, with a public comment period open until May 6, 2026. This regulation is distinct from the December 2025 draft and defines digital virtual humans as virtual figures existing in nonphysical environments that simulate human appearance and behavior using computer graphics, digital image processing, and AI. The draft mandates continuous labeling on all virtual human content, requires explicit consent before using any individual's likeness, voice, or personal data to create a digital human, and prohibits using digital humans to bypass facial recognition, voice recognition, or other identity authentication mechanisms. The regulation bans providing virtual intimate relationships, including virtual family members or romantic partners, to minors, and prohibits services that may induce excessive spending, encourage harmful behavior, or negatively affect minors' physical and mental health. Content involving sexual innuendo, violence, horror, or discrimination is prohibited. Providers are encouraged to intervene and provide professional assistance when users exhibit suicidal or self-harming tendencies. The April 2026 regulation forms part of China's broader push, articulated in the 15th Five-Year Plan and the AI Plus initiative released in March 2026, to balance aggressive AI deployment across the economy with tightened governance, particularly regarding youth protection and personal data.
The VTuber industry beyond China has also experienced turbulence relevant to understanding the sector's structural vulnerabilities. In July 2025, VShojo, one of the three major global VTuber agencies, ceased operations following accusations of withheld streaming residuals, underscoring persistent labor and financial governance issues across the virtual talent industry.
What emerges from this landscape is a picture of remarkable technological ambition and cultural adoption running alongside deep tensions. Millions of Chinese users have formed genuine emotional bonds with virtual idols, AI companions, and avatar-mediated social identities, driven by the convergence of subcultural community, platform design, generational demographics, and the deliberate engineering of emotional resonance by companies like Xiaoice and MiniMax. The commercial systems built around these bonds have generated enormous economic value but have also produced documented harms, from exploitative labor practices behind motion-capture performances to compulsive spending in fan economies to the psychological risks of parasocial attachment to entities incapable of reciprocation. China's regulatory response has been extensive and accelerating, moving from broad content governance to increasingly specific rules targeting the emotional, financial, and identity-related dimensions of human interaction with digital beings. The April 2026 draft regulation on digital virtual humans represents the most targeted articulation yet of the state's intention to govern not merely the technology but the psychological and social relationships it produces.
[Apr 2026]