Animating the Belt and Road Through Digital Humans

The Belt and Road Initiative represents China’s most ambitious global infrastructure and connectivity campaign, extending economic, political, and cultural influence across Asia, Africa, Europe, and beyond. Since its announcement by Xi Jinping in 2013, the initiative has expanded from physical transport corridors and port investments into digital domains through a parallel effort known as the Digital Silk Road. Digital humans—AI-powered virtual beings capable of realistic human-like interaction through speech, facial expression, and gesture—have emerged in recent years as a technology with potential relevance to this broader campaign. Their development intersects with key concepts in international relations scholarship, including soft power (the ability to shape preferences through appeal and attraction rather than coercion), digital diplomacy, and what Joseph Nye has described as narrative power in global affairs. The relationship between digital humans and the Belt and Road Initiative, however, is far more commercially driven and analytically inferred than the existing literature tends to acknowledge. No Chinese government document explicitly connects digital human technology to BRI objectives. Where these technologies do intersect with BRI geography, they arrive primarily through cross-border e-commerce platforms targeting Southeast Asian markets, not through coordinated state diplomacy.

The Digital Silk Road was formally proposed by Xi Jinping at the 1st Belt and Road Forum for International Cooperation on May 14, 2017. Its conceptual foundation traces to the March 2015 joint publication by the National Development and Reform Commission, the Ministry of Foreign Affairs, and the Ministry of Commerce: Vision and Actions on Jointly Building Silk Road Economic Belt and 21st Century Maritime Silk Road, which called for accelerating information infrastructure construction and cross-border fiber-optic cables. The DSR’s operational scope centers on hard digital infrastructure: 5G networks, submarine cables, data centers, cloud computing, e-commerce platforms, mobile payment systems, and surveillance technology. By 2022, China had signed DSR-specific memoranda of understanding with 17 countries and established Silk Road E-Commerce bilateral cooperation mechanisms with 23 countries. At the 3rd BRI Forum in October 2023, China and 13 countries jointly issued the BRI Digital Economy International Cooperation Beijing Initiative. An estimated $79 billion has been spent on DSR projects overall, with 201 Chinese digital companies implementing 1,334 overseas projects. The 15th Five-Year Plan (2026–2030), approved by the National People’s Congress on March 12, 2026, devotes its Chapter 23 to high-quality co-building of BRI and mentions expanding cooperation in green development, AI, and the digital economy, but does not reference digital humans. Nor did the 14th Five-Year Plan. The DSR’s documented deliverables—telecommunications towers, undersea cables, cloud data centers, smart city platforms—do not include virtual beings in any policy document, forum communiqué, or State Council directive.

Government policy support for digital humans within China is substantial but oriented entirely toward domestic industry development. No State Council document specifically addresses digital humans. The closest national-level policy is the 14th Five-Year Plan for Digital Economy Development, issued January 12, 2022, which covers virtual reality and augmented reality broadly but never uses the term “digital human” (shuzi ren). Digital human-specific policy first emerged at the municipal level: Beijing’s Action Plan for Promoting Digital Human Industry Innovation and Development (2022–2025), issued in August 2022 by the Beijing Municipal Bureau of Economy and Information Technology, was China’s first dedicated digital human industry policy at any government level, targeting an industry scale exceeding 50 billion RMB by 2025. The first national-level document to explicitly mention digital humans multiple times was the Metaverse Industry Innovation Development Three-Year Action Plan (2023–2025), jointly issued by MIIT and four other ministries in September 2023. That plan calls for innovating digital human development tool components, cultivating a batch of well-known digital humans, and creating digital human benchmark products and brands—envisioning guides, customer service agents, and shopping assistants, all in domestic applications with no BRI dimension. On standards, the China Academy of Information and Communications Technology (CAICT) has been the primary driver of digital human standards work, drafting the Virtual Digital Human Indicator Requirements and Evaluation Methods framework under the China Communications Standards Association. Eighteen companies—including Tencent, Baidu, Alibaba Cloud, Huawei, iFlytek, and Silicon Intelligence (guiji zhineng)—have participated in CAICT’s digital human system capability evaluation.

China’s regulatory architecture for digital humans is the world’s most developed, comprising multiple layers of binding regulation. The Deep Synthesis Provisions, jointly issued by the Cyberspace Administration of China (CAC), MIIT, and the Ministry of Public Security, took effect January 10, 2023. Article 23 explicitly defines covered technology to include 3D reconstruction, digital simulation, and other technologies for generating or editing digital characters and virtual scenes. The provisions require conspicuous labeling of synthetic content, informed consent for biometric editing, and security assessments for services generating face or voice content. The Interim Measures for Generative AI Services, issued by seven agencies and effective August 15, 2023, constitute China’s first binding comprehensive regulation targeting generative AI; by December 2025, 748 generative AI services had completed mandatory filing. The AI Content Labeling Measures, published March 14, 2025 and effective September 1, 2025, establish a two-tier system: explicit labels for high-risk content and implicit metadata labels for all AI-generated content, accompanied by the national standard GB 45438-2025. Most recently, the CAC published a Draft Digital Human Service Regulation on April 4, 2026, defining digital humans as virtual figures that exist in nonphysical environments and simulate human appearance and behavior. It mandates prominent labeling, explicit consent before using anyone’s likeness or voice, and bans on digital humans bypassing identity verification or targeting minors with virtual intimate relationships. These regulations are, however, inward-facing. No mechanism requires Chinese companies to apply domestic AI content regulations when deploying digital humans abroad. China’s promotion of its AI governance frameworks internationally occurs through the Global AI Governance Initiative, proposed in October 2023, and multilateral forums, not through regulatory extension to BRI partner countries.

The Chinese digital human industry has grown rapidly in commercial scale. Emergen Research valued China’s digital human avatar market at USD 418 million in 2024, projected to reach $4.1 billion by 2034. IDC predicts the broader AI digital human market will reach 250.5 billion yuan by 2029. China hosts between 1,200 and 1,500 active digital human companies, concentrated in Guangdong, Zhejiang, Beijing, and Shanghai. The firms producing these technologies span a spectrum from platform companies to specialized startups. Silicon Intelligence (guiji zhineng), headquartered in Nanjing and founded in August 2017, holds 32.2 percent market share in China’s digital human intelligent agent segment by revenue. Its flagship products include the DUIX ONE multimodal large model and the open-source HeyGem framework, which has accumulated over 13,000 GitHub stars. The company has deployed more than 80,000 digital human agents across 40 industries and 40,000 enterprises, with revenue growing from 223 million RMB in 2022 to 655 million RMB in 2024. It filed a Hong Kong Stock Exchange prospectus in October 2025 aiming to become the first dedicated digital human IPO. Baidu’s XiLing platform, launched December 2021, achieved a dramatic cost reduction with its September 2024 release cutting 3D hyper-realistic digital human creation from approximately 10,000 RMB to 199 RMB. MoFa (Shanghai) Information Technology Co., Ltd. (mofa (shanghai) xinxi keji youxian gongsi), founded in October 2017 and focused on 3D digital humans, has raised a cumulative $160 million with investors including ByteDance and SoftBank Vision Fund, was named a Gartner Cool Vendor in GenAI Native Applications in 2024, and showcased 3D digital human AIGC technology at Gitex Dubai 2024 alongside Alibaba. SenseTime’s SenseAvatar platform supports over 100 countries’ languages for text generation and voice synthesis, with 300 multilingual voice models. Xiaoice, spun off from Microsoft in July 2020, operates in China, Japan (as Rinna), and Indonesia. ByteDance released OmniHuman-1 in February 2025, generating full-body realistic video from a single image plus audio, and the open-source LatentSync 1.5 lip-sync framework in March 2025.

Among all digital human companies, iFlytek (keda xunfei) has the strongest verified connection to BRI-adjacent activity. At the 22nd China-ASEAN Expo in September 2025, iFlytek served as Multi-language Translation Service and AI Special Partner, deploying two digital human MCs at the opening ceremony—the first-ever use of digital humans at a CAEXPO event—alongside seven-language real-time AI simultaneous interpretation. It released the Spark ASEAN Multi-language Large Model supporting over 130 languages, including Arabic, Malay, Indonesian, Thai, and Vietnamese, and operates servers in Singapore, Saudi Arabia, and Ireland. This deployment represents the closest verifiable intersection of a specific digital human product with a named BRI-framework event, though it was a commercial partnership rather than a policy-directed deployment.

The most commercially active intersection of digital humans with BRI geography occurs through cross-border e-commerce livestreaming. Alibaba International Station launched a digital human livestream registration system on January 17, 2024, requiring Chinese Gold Suppliers to use Alibaba-approved digital human service providers and display Digital Human Live labels. BocaLive, developed by Shanjian Intelligence (shanjian zhineng) and launched March 21, 2024, operates as the first cross-border digital human livestream control system, supporting simultaneous deployment on Shopee, Amazon, TikTok, Lazada, AliExpress, Facebook, YouTube, Tokopedia, and Coupang in 12 languages including Malay, Indonesian, Thai, Vietnamese, Burmese, and Hindi. Documented cases include a Jiangsu Nantong seller gaining over 2,000 monthly orders on Shopee for children’s clothing, a Changsha seller reaching 3 billion Indonesian rupiah GMV in a single TikTok session in Indonesia, and a Shenzhen jewelry team achieving over 20,000 Thai baht per session on Thailand TikTok. Baidu’s HuiBoXing (huiboxing) platform announced 100,000 free digital human licenses for merchants in 2024 and expanded into Brazil via Shopee and Lazada in 2025. MoFa (Shanghai) Information Technology Co., Ltd. won the Cross-border Ecosystem Cooperation Partner Award from the Xiamen Cross-border E-Commerce Association in September 2024 for its multilingual 3D digital human solutions. These deployments target markets that happen to overlap geographically with BRI corridors, but they are driven by commercial logic—reduced labor costs, 24-hour multilingual operation, and scalable content production—rather than by any state directive linking digital human technology to the Belt and Road Initiative. Industry reporting from late 2024 and early 2025 also indicates a cooling trend, with platforms algorithmically detecting digital human streams and potentially restricting or penalizing them.

At physical trade events, digital humans have appeared at the Global Digital Trade Expo in Hangzhou, co-hosted by the Zhejiang Provincial Government and the Ministry of Commerce. The 2nd edition (November 23–27, 2023) gathered 50 global large AI models and featured a Silk Road E-Commerce VR pavilion showcasing 30 countries’ cultures. The 3rd edition (September 25–29, 2024) hosted the first-ever Digital Human Debate Competition, where eight large-model-powered digital humans competed in structured debates on AI applications across education, cultural tourism, livestream e-commerce, and healthcare. Within the Silk Road E-Commerce Pavilion at that event, Yiwu Small Commodity City’s Chinagoods platform displayed its Digital Boss Lady—a digital human clone of a Yiwu merchant offering 24-hour multilingual product presentations. The Chinagoods AI platform had by then covered 18,000 merchants, built AI-generated independent e-commerce sites for over 7,000 merchants, attracted 640,000 new clients, and driven trade volume growth exceeding 15 percent, with Kazakhstan and Thailand as guest countries of honor. The 4th edition (September 25–29, 2025) drew 1,812 exhibitors from 154 countries and over 42,000 professional buyers. These expos serve as showcases where digital human technology is exhibited alongside BRI commerce platforms, though they function as commercial trade fairs rather than instruments of directed state diplomacy.

In cultural diplomacy, the strongest verified case of a digital human in BRI-adjacent content involves Tianyu (tianyu), an ultra-realistic virtual being inspired by Dunhuang flying apsaras, created by Yuanyuan Technology (yuanyuan keji), a subsidiary of Tianyu Digital Technology. Tianyu performed at the 2023 Silk Road Spring Festival Gala on January 22, 2023, broadcast by Shaanxi Radio and Television Group under the guidance of the State Council Information Office and the National Radio and Television Administration. She performed Qingpingle alongside live performers, incorporating Yaozhou celadon, a national-level intangible cultural heritage item. The broadcast reached over 110 official accounts across Chinese platforms and overseas media, generating 18.8 billion media impressions and 46 trending topic appearances. Tianyu has accumulated over 5.5 million followers across Chinese social media platforms. Xinhua’s AI news anchors—Xin Xiaohao (launched November 2018, developed with Sogou), Xin Xiaomeng (March 2019), and the 3D anchor Xin Xiaowei (May 2020)—have multilingual capabilities in Chinese, English, and Russian. By May 2020, Xin Xiaohao had produced over 13,000 news reports totaling more than 35,000 minutes. Xinhua operates the Belt and Road Portal and the Xinhua Silk Road information service, but no specific evidence was found of these AI anchors being deployed for BRI-themed content. People’s Daily launched AI digital anchor Ren Xiaorong (ren xiaorong) in March 2023, supporting Mandarin, Cantonese, and English. These state media digital humans represent potential BRI communication infrastructure, but their actual deployment in BRI contexts remains undocumented as of April 2026.

In education, the primary BRI vocational training brand is the Luban Workshop (luban gongfang) program, which operates in 36 countries as of September 2025. Technology deployments within these workshops involve VR simulation laboratories and smart classrooms rather than digital human AI instructors. The Kyrgyzstan Luban Workshop, inaugurated in June 2024 at Kyrgyz National Technical University, includes a virtual simulation laboratory for water resources and electrical power training with over 300 items of equipment. The Thailand Luban Workshop at Ayutthaya Technical College uses VR and metaverse technology for cross-border teaching; in September 2023, Xinhua documented students in Thailand and China simultaneously wearing VR headsets for joint automotive parts assembly training. The Kenya Luban Workshop, co-built with Huawei, focuses on AI, cloud computing, and cybersecurity. These are immersive simulation environments, not AI avatar teachers, and the distinction matters for any accurate assessment of how digital technology is deployed within BRI educational programming.

Within China’s domestic governance apparatus, digital human deployments in public services are extensive and well-documented. Shenzhen’s Futian District launched 70 AI digital civil servants based on DeepSeek in February 2025, covering 240 government service scenarios. Wuxi deployed over 20 government digital humans in April 2025. Nanjing’s DeepSeek-based enterprise digital human customer service processes more than 100,000 daily inquiries with 92 percent accuracy. These deployments demonstrate the technological capacity that could theoretically be extended to BRI international cooperation zones, but no evidence of such extension was found. The China-ASEAN Information Harbor in Guangxi exists as a digital platform, and the China-Laos AI Innovation Cooperation Center involves customs automation, but neither has documented digital human deployments.

The geopolitical and cultural implications of Chinese digital technology exports along BRI corridors have generated a substantial body of critical scholarship, though this literature does not specifically address digital humans. Recorded Future’s 2021 report on China’s digital presence along the Digital Silk Road presents eight case studies from Africa, Latin America, and Southwest Asia, arguing that DSR projects in least-developed regions create power imbalances and cybersecurity risks. A 2020 study published through SciELO South Africa documents Chinese officials advising African government counterparts to adopt internet sovereignty frameworks through seminars and official visits. The European Parliament Research Service’s 2024 analysis warns that DSR infrastructure gives China leverage to advance governance models that may not align with liberal democratic norms. Counter-perspectives deserve equal weight. An April 2026 analysis in The Diplomat argues that the digital colonialism narrative is oversimplified, noting that African states exercise considerable agency in shaping digital governance and actively leverage competing vendors. Forrester research documents ASEAN nations pursuing deliberate diversification across local, American, and Chinese cloud providers. Harvard’s Davis Center notes that Kazakhstan developed its own Digital Kazakhstan 2020 strategy partly as a counterweight to DSR influence. The gap in this literature is significant: no academic paper has specifically examined Chinese digital human deployments in BRI partner countries. The academic conversation about digital humans remains focused on domestic Chinese consumer markets, fan culture, and virtual idol identity formation. Any framing of digital humans as instruments of digital colonialism in BRI contexts is currently speculative extrapolation from the broader digital infrastructure literature, not an evidence-based finding.

Several challenges attend the further development and potential internationalization of digital humans as tools of engagement. Technological access and infrastructure remain uneven among BRI participants, with many partner countries lacking the bandwidth, computing resources, and platform ecosystems needed to deploy or consume digital human content at scale. The cost of producing high-quality 3D digital humans has fallen dramatically—Baidu’s reduction to 199 RMB per avatar represents a significant democratization—but the infrastructure to run interactive, real-time digital humans requires stable cloud computing access that is not universally available. Ethical concerns arise around the authenticity of AI-mediated interactions, particularly in educational and diplomatic contexts where the distinction between human and synthetic communication carries normative weight. Privacy and surveillance risks are amplified by data-driven avatars, especially in regions lacking robust digital governance frameworks. China’s own regulatory trajectory, culminating in the April 2026 Draft Digital Human Service Regulation, suggests that Chinese regulators themselves view digital humans as requiring oversight to prevent misuse—including potential emotional manipulation of users and unauthorized appropriation of individuals’ likenesses.

The relationship between digital humans and the Belt and Road Initiative is best understood as an emergent commercial phenomenon that overlaps geographically with BRI corridors, rather than as a policy-directed deployment of technological soft power. Chinese digital human companies are expanding internationally because Southeast Asian and Middle Eastern markets offer growth opportunities in e-commerce, customer service, and content creation, and the firms best positioned to serve those markets happen to be Chinese. iFlytek’s digital human MCs at the China-ASEAN Expo, Tianyu’s performance at the Silk Road Spring Festival Gala, and the cross-border e-commerce avatars operating on Shopee and TikTok across Indonesia, Thailand, and Vietnam represent the verified points of contact between digital human technology and the BRI’s geographic and institutional footprint. State media organizations like Xinhua and People’s Daily have built AI anchor capabilities that could in principle be channeled into BRI-themed content production, but no such deployment has been documented. For researchers and policymakers, the more analytically productive frame is not whether China is using digital humans as instruments of Belt and Road influence, but whether the commercial diffusion of Chinese digital human platforms into BRI partner countries will produce the same kinds of data dependency, content sovereignty, and governance norm questions that have attended the broader Digital Silk Road. As AI capabilities continue to improve and the cost of producing convincing digital humans continues to fall, these questions will become more urgent regardless of whether any Chinese policy document ever explicitly instructs a digital human to speak on behalf of the Belt and Road.