Subscribe to The Informer for monthly expert analysis, and to Events for advance notice of visiting world leaders and distinguished guests.
You may unsubscribe from Lowy Institute newsletters at any time. For information on our privacy practices and how to unsubscribe, see our Privacy Policy.
The most-pressing world events explained by Lowy Institute experts and global contributors, in your inbox, every Thursday.
You may unsubscribe from The Interpreter at any time. For information on our privacy practices and how to unsubscribe, see our Privacy Policy.
Artificial intelligence, explained.
A staff member works on script refinement, on-location filming, AI post-production, and overseas translation and distribution at a micro drama production company on 21 July 2026 in Jiaxing, China (VCG via Getty Images)
A face licensed for one AI advert can be reused indefinitely, and local laws have not caught up with that.
Losing face? Licensing and governing AI content in ChinaLosing face? Licensing and governing AI content in China
Would you accept US$15 to let an AI company use your face in artificial intelligence generated content?
In China, this is becoming a real proposition. A recent Rest of World (Opens in new window) article reports that people are paid from between US$15 to US$700 to license their facial likenesses for AI-generated microdramas and advertisements produced for mobile phone viewing.
China’s mobile entertainment industry is already enormous – and the use of AI-generated content is growing. More than 95% of microdramas released in China during the first three months of 2026 reportedly used AI somewhere in their production. The industry needs more faces – celebrities and ordinary people – to meet the demand for fast-paced, audience-driven microdrama production (Opens in new window).
One response has been straightforward: pay people for their faces.
At the same time, regulators and platforms are cracking down on harmful and pirated content (Opens in new window),unauthorised digital likeness (Opens in new window), and homogeneous AI-generated characters.
Licensing might be “legal” rather than simply scraping somebody’s photographs or reproducing their likeness without permission. But it creates a new problem for AI governance. An agreement might authorise one advertisement. But could the same AI facial model later appear in 100 advertisements? A romance or a political drama? Could it be aged, sexualised or made to say things you would never say?
Generative AI transforms the question of facial rights. It is no longer only about protecting people from having their identity stolen. It is about governing what an authorised digital likeness can be made to do – potentially forever.
Who controls the image? (Wang Gang/China News Service/VCG via Getty Images)
A face can now operate simultaneously in at least three ways.
It remains a social identity, through which people recognise one another and negotiate reputation, recognition and belonging.
In Chinese history, the face has been understood through social relationships, to capture ideas of reputation, dignity, recognition and social standing. To “lose face” traditionally means (Opens in new window) to suffer a loss of status or social standing in the eyes of others.
In the digital era, the face has become a biometric identifier. Facial recognition presents a difficult governance challenge, requiring adaptive policy (Opens in new window) response to disruptive technologies and changing citizen expectations (Opens in new window). It turns physical appearance into machine-readable information that can authenticate a payment, unlock a device, verify an identity for transaction or access to a physical space, and trigger surveillance networks in population management.
Generative AI is now turning the face into something else: a reproducible digital asset.
Facial recognition is primarily concerned with identification, “who is this person?” Generative AI asks, “what can this face do?” The face is detachable from the body, reusable across time and space, through which different stories and formats can be produced for different audiences across platforms. This raises questions about consent and ownership when a face can be licensed for AI generative images.
Who shares in the ongoing value – and who bears the reputational cost – of an indefinitely productive likeness.
China already has substantial laws governing faces. Under the Civil Code, portrait rights form part of a person’s broader personality rights. Individuals have rights over the production, use and publication of their likeness, while unauthorised fabrication or misuse through information technology can constitute an infringement.
Biometric information is also treated as sensitive personal information under China’s personal information protection regime. The 2025 Measures on the Security Management of Facial Recognition Technology Applications (Opens in new window), which came into force on 1 June 2025, define facial information as biometric information associated with an identified or identifiable natural person and establish requirements for the processing and protection of such information.
China also regulates generative AI. The Interim Measures for the Management of Generative Artificial Intelligence Services (Opens in new window), effective since August 2023, cover generative services producing text, images, audio and video and explicitly connect AI governance with personal information, data security and the protection of individual rights.
Yet China’s emerging facial economy raises a governance question that goes beyond privacy and rights. It is also about value: when a face becomes a reusable AI asset, its value is not exhausted by the initial licensing fee. A modest one-off payment may enable repeated commercial use across advertisements, dramas and platforms, while the resulting value rests with AI firms, studios and distributors.
The question is therefore not only whether a person consented, but who shares in the ongoing value – and who bears the reputational cost – of an indefinitely productive likeness.
Licensing could offer people greater control over their likeness. But consent is not the end of the story. It should not become a one-time transaction for an indefinitely reproducible facial identity whose future uses cannot be anticipated.
This challenge cuts across portrait and personality rights, data protection, contract law, labour and the distribution of economic value. A meaningful licensing framework should specify permitted uses, duration and scope. It should also prohibit sensitive uses, enable traceability, and consider whether continued commercial use requires renewed consent or further payment.
The challenge for licensing and governing the face is not simply to prevent identity theft. It is to determine how real human faces can be governed when they become raw material for synthetic worlds, from consent to taking control over where they go, what they do, and what value they generate.
About the authors
Haiqing Yu
Haiqing Yu is Professor of Media and Communication and Australian Research Council (ARC) Future Fellow at RMIT University, Melbourne, Australia.
Joanne Kuai
Joanne Kuai is a Research Fellow at the School of Media and Communication, RMIT University, Australia, and holds a PhD from Karlstad University, Sweden.
