Published 49 minutes ago • loading... • Updated 1 hour ago
U.S. Warns China Is Narrowing AI Gap as Model-Risk Debate Grows
Chinese startups are releasing models that challenge U.S. leaders, while American agencies say some developers copied frontier systems.
On September 8, the FBI, National Security Agency, and Cybersecurity and Infrastructure Security Agency accused Chinese developers of "industrial-scale distillation," claiming firms extracted capabilities from advanced U.S. AI models through targeted activities.
The U.S. has outpaced other nations in AI research for years, but experts report China is rapidly catching up this year, appearing to erase the established American lead despite U.S.-led restrictions.
Beijing-Based startup Moonshot released Kimi K3, while Z.ai launched GLM-5.3 in August, raising Silicon Valley eyebrows; DeepSeek previewed its powerful V4 model featuring improved "agentic" capabilities.
Treasury Secretary Scott Bessent stated Chinese developers "distill our models and they can never get ahead of us," while China's state security minister Chen Yixin warned AI could threaten national security if used by "hostile forces."
This debate intensifies ahead of Chinese President Xi Jinping's expected visit to Washington this month to meet President Donald Trump, as experts say the future of AI depends on establishing global governance frameworks.