‘Silent Cold War’: Why calls to slow AI have sparked new US–China frontier

What the report says

Anthropic chief executive Dario Amodei has renewed debate over AI safety by calling for the most powerful systems to be developed more slowly and for tougher limits on exports of advanced AI chips to China. In comments reported by Al Jazeera, Amodei warned that rapid progress in frontier models could create severe security and societal risks, and he argued that Washington should keep tighter controls on cutting-edge technology reaching Chinese firms.

The proposal drew an immediate response from Beijing. China’s Foreign Ministry said countries should take an open and inclusive approach to artificial intelligence and warned that confrontation and rivalry would undermine global AI governance. A state-run Chinese newspaper went further, portraying the call as part of a Cold War-style effort to curb China’s AI development and preserve US technological dominance.

The dispute comes amid a broader US-China race over AI capability, funding and hardware. Al Jazeera noted that American companies still lead in investment and in access to the most advanced chips used to train frontier models, while Chinese firms have shown they can still build competitive systems despite export restrictions. The issue matters because both governments view AI as strategically important for economic power, national security and military use, making safety arguments inseparable from industrial policy.

For context, this debate is unfolding alongside wider export-control fights over semiconductors, rare earths and cloud infrastructure. It also reflects a growing tension in Washington and Beijing between calls to manage AI risks and efforts to secure a national advantage in a technology that could shape future military, intelligence and commercial competition.