Shenzhen University Professor Proposes Risk-Based AI Governance Framework

Associate Professor Cao Jianfeng of Shenzhen University has proposed a risk-based governance framework to manage the proliferation of AI algorithmic decision-making. The framework emphasizes transparency and traceability, specifically for high-risk applications in lending, recruitment, and healthcare, to prevent deception and algorithmic discrimination.

Cao advocates for the principle of value alignment to ensure AI behavior remains consistent with human ethical principles. He notes that the Standardization Administration of China's National Technical Committee 260 on Cybersecurity integrated this concept into the AI Safety Governance Framework 2.0 in September 2025. A central tenet of his proposal is the human-in-the-loop principle, which asserts that decisions affecting major rights, liberty, and life must ultimately be made by humans.

To incentivize safety over speed, Cao proposes reforming product liability systems. He argues that holding AI developers and providers financially accountable for defects would force businesses to prioritize safety. These proposals align with broader efforts by the Government of China to enhance safety oversight and risk monitoring through specific AI provisions within its Cybersecurity Law.