•  
  •  
 

UC Law SF International Law Review

Abstract

Since 2021, China has strategically expedited artificial intelligence (AI) legislation and refined algorithmic governance to compete with the US and the EU. The rapid growth of algorithm filing cases is a direct result of the “Made in China 2025” Initiative which prioritizes AI development driven by domestic leading tech companies like DeepSeek. As world-class digital powers, China, the US, and the EU diverge in their categorization of AI risks and adoption of regulatory measures. China’s centralized single-agency regulatory infrastructure differs from the multi-agency and decentralized governance models in the US and the EU, respectively. China does not impose the same level of obligations on AI deployers as on service providers, rarely punishes digital platforms, and prefers local laws and judges in dispute resolution. We argue that China should learn from the EU and the US and encourage the AI industry to self-regulate and delegate more powers to local regulators and entities. Regarding transparency, China’s AI filing system focuses on AI systems with public opinion influence, while the US and the EU require high-risk systems to meet reporting and certification obligations in a non-ideological manner. All three jurisdictions have employed sophisticated AI detection and labeling systems, but user notification and stakeholder protection in China are not as mature. Regarding law enforcement, China focuses on regulating AI operators instead of protecting users. China could benefit from the EU’s approach by refining rules to include a broader range of stakeholders, such as deployers and their employees. Regarding security assessment, the US and the EU emphasize continuous risk management post-deployment, while China integrates ex-ante assessments with national security considerations. Their security assessments have formal and functional divergences in mandatory nature, effect, and methods due to different legal traditions and sociopolitical contexts. AI has been widely applied in specific fields like content moderation and judicial reasoning. All three jurisdictions have strikingly different regulatory philosophies, focuses, and methods for AI content moderation. We argue that China should learn from the EU to avoid the unacceptable risks of judicial AI.

Download

Share

COinS