China's AI Security Focus Shifts to Cyber Threats

Chinese policymakers are increasingly viewing advanced AI models as a national security threat due to their alarming hacking capabilities, shifting focus from immediate risks like deepfakes to broader cybersecurity dangers. The country's spy agency chief warned this week that the technology could jeopardize national security. A recent demonstration showed a small California team using AI to build a cyberattack "worm" that could compromise millions of accounts on China's mega-app WeChat, highlighting potential vulnerabilities. Experts note China and the US share similar AI safety concerns, though these issues are less publicly debated in China, with cybersecurity impacts of autonomous AI agents ranking high among government priorities. The topic is expected to feature in talks between President Xi Jinping and US counterpart Donald Trump on Thursday, despite their rivalry in the field. Both nations are developing a shared understanding of risks, though cooperation on safety evaluations may be undermined by competition, with China wary that "security" could be used to restrict its technological development.