º¬Ðß²ÝÉçÇø

Artificial intelligence-generated content (AIGC) is sparking a global transformation, presenting China¡¯s AI sector with an unprecedented opportunity for rapid expansion. A growing body of sector-specific regulations has shaped an evolving legal oversight framework for AI. At the same time, a rise in AI-related disputes is enriching judicial practice with valuable case references.

This article outlines the core regulatory requirements for AIGC platforms under current Chinese law and, through analysis of recent cases, focuses on the platforms¡¯ duty of care ¨C offering valuable legal guidance for compliance-oriented operations.

Regulations

Today, China¡¯s legislative approach to AI is shifting from framework design to practical implementation. Anchored by the Cybersecurity Law, Data Security Law and Personal Information Protection Law, the regime now includes AI-specific conduct rules and compliance duties articulated through administrative regulations, departmental rules and other normative instruments. Regulatory expectations centre on three core areas: (1) data security and personal information protection; (2) content safety and intellectual property rights; and (3) algorithmic fairness, transparency and tech ethics review. These requirements, guided by the principle of balancing development with security, translate into concrete obligations and standards of care for AIGC platforms across data, content and algorithmic operations.

Additional AI-related rules are in the pipeline, including the consultation draft on the Interim Measures for the Administration of Anthropomorphic Interactive AI Services. Lawmakers and regulators remain vigilant regarding the novel risks accompanying rapid AI advancement and adoption. By identifying and filling regulatory gaps, they are working towards a more integrated, adaptive and robust system of legal governance for AI in China.

Duty of care

Unlike conventional internet service providers, AIGC platforms perform a dual function, generating content via AI tools while simultaneously offering technical services. This dual character has led the courts to apply stricter standards of care, supplementing the obligations already set out in legislation.

Under recent judicial cases, the decisive question in establishing whether an AIGC platform bears liability for infringement is whether it has discharged its reasonable duty of care. The judicial test considers whether an ordinary service provider in the field would have been unlikely to spot the infringement, or whether the platform can show it adopted all technically feasible precautions available at the time, which nonetheless failed to prevent the damage. This assessment determines whether the duty of care was met and, in turn, whether liability attaches.

Drawing on current legislation and recent judicial rulings, AIGC service providers are expected to observe the following duties of care.

1. 1. Platforms must exercise stringent oversight to identify content that is toxic, harmful or unlawful under current legislation. This emphasises the implementation of robust review processes rather than the absolute prevention of such material. Regarding inaccuracies that do not fall within these statutory prohibitions, platforms are not required to certify the veracity of the information.
   2. Platforms must deploy suitable and conspicuous warnings to ensure user awareness of functional constraints. Service contracts or analogous instruments should likewise be used to prohibit rights infringements, serving as an effective warning and precautionary measure.
   3. Reasonable scrutiny must be applied by platforms to assess whether featured or high-visibility positions display or promote infringing material. Stricter obligations apply to services, including reusable models and AI avatars, where the likelihood of infringement is pronounced, and the risk of wider distribution is elevated.
   4. A fundamental duty of care applies to the reliability of platform functions, requiring the implementation of prevailing technical measures, consistent with the state of the art, to improve the precision of content outputs.
   5. Platforms are required by law to conspicuously label AI-generated content. This obligation respects the public¡¯s right to know and serves as a protective function for rights holders, enabling them to recognise content as AI-generated and pursue more precise remedial strategies to safeguard their interests.
   6. Platforms must establish robust complaints and reporting mechanisms, enabling timely technical intervention to restrict and remove infringing content.

In light of the above-mentioned regulatory landscape, AIGC platforms are required to harmonise technological progress with robust compliance across data, model security, content and IP, and general care obligations. The development of an evolving internal compliance architecture is critical to striking the right balance between innovation and regulatory conformity.

Quan Wei is a partner and Zhu Wenhan is an associate at Jingtian & Gongcheng