China Advances AI While West Debates Safety Risks

The U.S.-China AI race is intensifying as Chinese models close performance gaps with American rivals, and both governments warn of security risks from the technology they are racing to build.

• No. 3 globally in July, China's Kimi K3 has since fallen to ninth place.
• Stanford's Human-Centered AI institute found China appears to be erasing the U.S. lead this year.
• Sept. 8: FBI, NSA, and CISA accused Chinese developers of extracting restricted U.S. model capabilities.
• China rejected the accusations and called U.S. claims an attempt at an AI industry monopoly.

Why it matters: Both countries frame AI as a national security issue, yet competition and distrust make cooperation difficult, according to researchers at the Harvard Belfer Center and Fudan University.

• China's state security minister Chen Yixin warned AI could be used to spread political misinformation and attack critical infrastructure.
• Lizzi C. Lee of the Asia Society Policy Institute said the U.S. may need to treat frontier AI outputs, not just chips, as a form of technology transfer.

How 21 sources split on this story

Where they split: The central dispute is whether calls to slow AI development are legitimate safety policy or, as Chinese officials argue, a strategy to block China's rise.

What’s next: Xi is expected in Washington this month for a summit with Trump where AI is a primary topic.

• Trump and Xi agreed in May to hold high-level AI talks; none have been publicly scheduled.
• Will the Trump-Xi summit produce any binding AI governance commitments, or remain aspirational?
• Can U.S. export controls on chips and model outputs meaningfully slow China's AI progress, given open-weight model proliferation?
• How will the U.S. verify or contest the distillation allegations it has leveled against Chinese developers?