China is advancing legislation on the full lifecycle of AI: enterprises are required to more proactively disclose risks, and a regulatory sandbox will also be established for trial-and-error.

Beating AI News, November 3 — China’s National Development and Reform Commission (NDRC) announced it will accelerate comprehensive legislation on artificial intelligence (AI), establishing safety standards and ethical guidelines covering the entire lifecycle of AI from R&D and application, spanning data, computing power, algorithms and specific use cases. AI developers and operators will be required to disclose safety risks in a timely manner and proactively report incidents and vulnerabilities. While tightening regulation, the policy will also leave room for enterprises to test and iterate. The new rules will set a reasonable transition period, and support industries to establish "regulatory sandboxes" — controlled environments allowing enterprises to test new technologies. The sandboxes will come with dedicated application, testing, use and exit mechanisms, with regulatory rules subject to regular evaluation and timely adjustment. Beyond model and technical safety, regulation will also focus on AI’s impact on individuals, including monitoring risks of AI displacing jobs, improving social security systems, and preventing user addiction to AI and emotional dependence. Deepfakes, algorithmic discrimination and privacy breaches are also included in the governance scope. China introduced public-facing measures for generative AI services in 2023, and added comprehensive AI legislation to the State Council’s work plan in May this year. Regulatory sandboxes have already been piloted in some AI service sectors.