OpenAI Chief Executive Officer Sam Altman warned that Washington is misjudging the speed and breadth of China’s artificial-intelligence progress, saying Beijing-backed labs could outpace the United States in deploying large-scale inference systems even if American firms still lead in frontier research. “I’m worried about China,” Altman told reporters in San Francisco on 18 August, adding that the contest involves multiple layers—research, products and production capacity—rather than a simple technology leaderboard. Altman argued that U.S. semiconductor export controls, which now block most cutting-edge GPUs but carve out a ‘China-safe’ category while forcing Nvidia and AMD to remit 15 % of China sales to the federal government, are unlikely to arrest China’s momentum. Even a total ban, he said, could be sidestepped through black-market channels or domestic fabrication. His remarks echo earlier scepticism from Nvidia CEO Jensen Huang about the policy’s effectiveness. Competition from Chinese open-source models also shaped OpenAI’s recent decision to release two open-weight language models, gpt-oss-120b and gpt-oss-20b. “If we didn’t do it, the world was gonna be mostly built on Chinese open-source models,” Altman said, underscoring how China’s developers such as DeepSeek and Kimi K2 are pressuring U.S. firms to widen access to their own technology. Altman’s comments come as Beijing steps up its own AI-safety regime, issuing national standards and mandating pre-deployment risk reviews. Analysts say that while China is tightening controls at home, the United States has not held formal bilateral talks on AI risks since 2024, raising questions over whether the two powers can coordinate on security while competing over technological leadership.