When Washington and Beijing agree on something, pay attention. Following urgent calls from Anthropic’s Dario Amodei, OpenAI’s Sam Altman, and Elon Musk to pause frontier AI development, both the US and China have officially rejected the idea. That’s a remarkable alignment, even if it happened for completely different reasons.
China’s Foreign Ministry was direct about its position. As reported by Android Headlines, spokesperson Guo Jiakun told reporters at a Beijing press briefing that ‘fearmongering, confrontation, and vicious competition will only disrupt the process of global AI governance.’ The message was clear: China sees the safety warnings not as genuine concern, but as political strategy.
State-run outlet Global Times went further, publishing an editorial accusing Silicon Valley executives of using safety rhetoric to protect their own market position. The paper argued that proposals to restrict chip exports and limit Chinese AI progress were ‘packed with containment provisions’ and amounted to a new Cold War playbook. Beijing also pushed back on US cybersecurity agencies’ claims that Chinese labs systematically copied capabilities from models like Claude and GPT-4, with China’s Commerce Ministry calling those accusations groundless and noting that model distillation is standard practice industry-wide.
This matters because it exposes a structural problem with any international AI governance effort. The two nations most capable of building frontier models have now both publicly ruled out a slowdown. President Trump made his position clear while speaking to reporters in Ireland, stating that ‘whoever wins AI, wins.’ That framing treats AI as a geopolitical race, not a technology to be managed carefully. Beijing’s framing is different in language but identical in outcome.
So the AI safety community, which already struggles to get traction inside individual companies, now faces a situation where the two dominant AI powers are actively competing against each other and dismissing slowdown proposals as interference. Companies like Anthropic and OpenAI have built entire policy teams around the idea that governments can be persuaded to regulate carefully. That theory just took a significant hit.
Still, neither government is completely ignoring the issue. Chinese State Security Minister Chen Yixin recently called for faster systems to manage AI security risks, and Xi Jinping continues promoting open-source AI frameworks for developing nations. Trump and Xi are scheduled to meet on September 24 to discuss AI governance, which could produce something concrete or nothing at all.
But the baseline position from both sides is now on the record. No pause, no moratorium, no coordinated slowdown. For developers building on top of frontier models, that means the models keep getting bigger and more capable. For anyone betting on international safety agreements to shape the next generation of AI, this week was not a good one.
