China’s AI Realism Is Not a Conspiracy

Summary

Fred Gao, writing with CT Zhao and X.PIN, offers a counterargument to ChinaTalk’s account of why Chinese AI researchers distrust Dario Amodei and American AI leaders. He argues that describing Chinese views as conspiracy theories or propaganda applies an uneven standard: American leaders’ beliefs are treated as sincere political convictions, while Chinese researchers’ interpretations of US behavior are dismissed. In Gao’s view, Chinese researchers are applying a realist framework in which great powers use technology, institutions, alliances, and rules to preserve strategic advantages. Their conclusions, he says, also reflect direct experience: many have studied or worked in the United States, followed frontier research, collaborated with American colleagues, and observed recent changes in US policy firsthand. The essay focuses on two positions it attributes to Dario: restricting China’s access to AI chips, models, and talent, while also seeking dialogue on a shared US-China regulatory framework. Gao says these aims may be sincerely motivated but appear contradictory to Chinese audiences, especially when paired with criticism of China’s political system. From that perspective, AI safety language can be understood as an extension of older power relations and restrictions on China’s development. He argues that negotiation cannot depend on either side accepting the other’s moral self-description or ideological narrative. Instead, the US and China should acknowledge mutual security concerns and competitive interests, narrow cooperation to specific risks such as AI-enabled cyberattacks, military miscalculation, and loss-of-control incidents, and establish reciprocal, verifiable, enforceable rules. In his view, institutions that can function without mutual trust are more realistic than cooperation built on assumed goodwill.