China rejects AI ‘threat narratives’ as US rivalry shapes safety debate

Beijing answers the slowdown debate

China’s Foreign Ministry rejected what it described as narratives of threat, confrontation and malicious competition in artificial-intelligence governance on September 14. Spokesperson Guo Jiakun said governments should pursue AI development that is open, inclusive and broadly beneficial. The intervention came as the United States debated calls from senior technology executives and researchers to slow the development of increasingly capable systems.

The immediate dispute is not simply whether advanced AI presents risks. It concerns whether a slowdown could be coordinated without giving one country a strategic advantage. Anthropic chief executive Dario Amodei had argued that development should be paced more cautiously while preventing China from moving ahead. President Donald Trump rejected a slowdown and framed the issue as a race the United States must win.

Beijing’s response highlights the diplomatic obstacle at the centre of AI safety policy. Washington and Beijing may share concerns about loss of control, cyber misuse and national-security applications, yet each fears that binding limits could leave the other side with stronger models, more computing capacity or greater influence over technical standards. Mutual suspicion makes verification and reciprocal restraint difficult.

Cooperation and competition coexist

The two governments nevertheless participated in the G20 Innovation Ministerial in North Carolina earlier in September. A consensus statement backed flexible, pro-innovation policy frameworks, trusted adoption, technical standards, workforce development and protection of intellectual property. The agreement demonstrates that broad principles are possible even while the countries disagree over regulation, export restrictions and the pace of frontier-model development.

China is also trying to shape institutions around open-source AI and international governance, while the United States’ official strategy explicitly treats leadership over China as economically and militarily important. These positions are not exact opposites: both promote innovation and acknowledge standards, but they assign different weight to national advantage, regulatory restraint and state oversight.

The next diplomatic test is whether AI enters senior US-China talks expected around Xi Jinping’s planned visit to Washington. Concrete progress would require more than general calls for cooperation. Negotiators would need to identify measurable subjects such as model evaluations, incident reporting, safeguards for biological and cyber capabilities, compute governance or emergency communication channels. Until then, the September exchange shows that AI-risk warnings are becoming part of strategic diplomacy rather than remaining a debate confined to technology companies.