China Updates AI Safety Framework Amid Western Tensions

The Cyberspace Administration of China recently updated its AI Safety Governance Framework to address the risks of recursive self-improvement and the potential for AI loss of control. While the Chinese government enforces strict model registration and content labeling, it currently lacks legal requirements for testing catastrophic risks, such as the creation of biological weapons.

Chinese officials and industry leaders have characterized Western warnings regarding AI existential risks as fear marketing intended to preserve United States market dominance. Despite this skepticism, senior officials have acknowledged new threats, noting that AI-powered cyberattacks have rendered traditional defenses obsolete.

Geopolitical tensions persist as Anthropic CEO Dario Amodei called for restrictions on China's access to advanced chips to ensure a democratic lead in the field. In contrast, Xi Jinping told President Trump that the two nations should manage AI for good, although mutual distrust continues to hinder the possibility of concrete safety agreements.