Recent discussions about ordinary individuals making 3,500 yuan in three days by selling their digital likeness, alongside the concept of AI image authorization, have sparked considerable public interest. As the trend moves from pioneering celebrities to open platforms for the general public, questions arise about the true nature of this business, its benefits, and its inherent risks. For those considering this path, understanding how to navigate the potential pitfalls is crucial.
How does AI image authorization work, and does the process differ between celebrities and regular people?
According to Li Ziqiang, head of AI business at Ciwen Media Co., Ltd., the primary method involves registering on relevant platforms, uploading a personal image, and signing an agreement. The system currently categorizes AI digital image authorization into two tiers: one for ordinary individuals and another for professional actors. Ordinary people are typically paid per minute of appearance in the final production, with rates ranging from a few hundred to a thousand yuan per minute. In contrast, professional actors negotiate on a case-by-case basis, with prices set by scope and duration, usually ranging from several thousand to tens of thousands of yuan. For the average person, the most immediate benefit is the low-cost realization of an on-screen dream, transforming their image into a sustainable digital asset. For the industry, this influx of content diversifies supply and pushes production towards more inclusive and intelligent methods.
Xu Xiaodan, a partner at Beijing Yuankun Law Firm and deputy director of the Beijing Lawyers Association's Culture, Tourism, Media, and Sports Committee, clarifies that portrait rights are a form of personality right and cannot be transferred or bought outright; they can only be licensed for use. Furthermore, facial features and voiceprints are classified as sensitive biometric information, requiring explicit, separate consent. Legally, celebrities and ordinary citizens hold equal rights; the differences lie in commercial value, social influence, and legal recourse capabilities. Ordinary individuals often lack professional teams to review contracts, making them more vulnerable and less capable of identifying risks in licensing agreements.
What are the positive aspects and potential risks associated with AI image authorization?
Li Ziqiang believes that collaboration between AI and real people brings vitality to the industry, viewing them as complementary rather than substitutive. The public can use their AI avatars for personal archival and family memories, as well as for small-scale monetization through roles in AI-generated films, short video creation, and virtual livestreaming. Celebrities, on the other hand, can overcome physical and scheduling constraints, extending their careers, revitalizing classic intellectual property, and achieving cost-efficiency in film reshoots and advertising. However, opportunities are accompanied by risks. Xu Xiaodan points out that some disreputable platforms offer low-priced, unequal contracts with hidden terms like 'perpetual authorization,' 'permission to sublicense,' or 'use for model training.' Signing such agreements could lead to the repeated circulation of one's facial data. Additionally, ordinary individuals face significant hurdles in legal recourse, as proving a specific AI character originates from their own facial model after multiple iterations involves high technical appraisal costs and dispersed responsibilities.
How can individuals protect themselves if they wish to try this, and what improvements are needed in the industry?
Xu Xiaodan advises the public to strictly choose compliant platforms with verifiable credibility and to understand all contract details thoroughly. When signing, it is crucial to reject vague terms such as 'perpetual authorization,' 'unlimited sublicensing,' or 'use of facial data for AI model training.' Individuals should strictly limit the authorization period and specify the project, ensure separate consent clauses are signed for facial features and voiceprints, and require the complete deletion of original materials, training models, and derivative data upon contract expiry, keeping proof of destruction. On the platform and enterprise side, she recommends encrypted storage of biometric data, auditable data destruction processes after authorization ends, and clear labeling of all AI-generated content. She calls on industry associations to promptly establish 'AI digital human authorization industry standards' to provide guidelines on authorization processes, data destruction, and dispute resolution, thereby enhancing legal risk awareness among market participants.
Li Ziqiang notes that the industry is actively promoting the establishment of a comprehensive system for portrait rights compliance, authorization confirmation, tiered licensing, and transparent revenue-sharing mechanisms to solidify the foundation for the sector's healthy, long-term development.