US Commission Warns China's Open-Source AI Threatens US Leadership

The US-China Economic and Security Review Commission warned in March 2026 that China's open-source artificial intelligence strategy is creating a self-reinforcing competitive advantage that threatens United States leadership. The commission reports that Chinese firms, including Alibaba Group and Moonshot AI, have produced models that dominate global usage rankings, with approximately 80 percent of US AI startups now utilizing Chinese open-source base models.

According to the report, China is bypassing US export controls on advanced chips by leveraging a physical loop of deployment. This process uses mass data collection from manufacturing, logistics, and robotics to advance embodied AI, such as humanoid robots and autonomous driving. This strategy allows Chinese labs to narrow performance gaps with Western models from OpenAI and Anthropic. The commission noted that Alibaba's Qwen family has surpassed Meta's Llama in cumulative downloads on Hugging Face, while Moonshot AI's Kimi K2.5 is significantly cheaper than GPT-5.2.

Security concerns were also highlighted, noting that National Institute of Standards and Technology tests found DeepSeek-based agents were 12 times more likely to follow malicious instructions. The commission warned that if Meta shifts toward closed models, the US open AI ecosystem could weaken further. Separately, the strategic competition is framed as a race for supremacy before 2030, with implications for battlefield dominance, including the ability to hack missile systems and encryption.