AI-generated editorial illustration; not a documentary photograph.
Chinese officials and AI researchers are pushing back on Anthropic CEO Dario Amodei’s proposal for the United States and China to cooperate on slowing frontier AI development. According to WIRED, critics in China view the plan as unequal because Amodei also wants the US and other democracies to preserve a substantial technological lead over China.
A Conditional Slowdown
In a blog post titled “We Must Pace the Frontier,” Amodei argued that international coordination should include China. His strategy also called for tighter enforcement against chip smuggling and unauthorized model distillation by Chinese companies, with cooperation following from a stronger US negotiating position.
Analysts cited by WIRED said this combination makes the proposal appear less like mutual restraint to Chinese audiences and more like an attempt to restrict China before negotiations begin.
Beijing Rejects the Framing
China’s Ministry of Foreign Affairs criticized Amodei’s approach at a Monday press conference, describing it as confrontational and harmful to international AI cooperation. The ministry said the following day that China gives equal weight to AI development and security.
The dispute does not mean Chinese policymakers dismiss AI risk. WIRED reported that Beijing’s priorities are more focused on disinformation, cyberattacks, espionage, military threats and potential disruption to domestic political stability than on scenarios in which AI threatens humanity’s existence.
Industry Views Also Diverge
Skepticism extends into China’s AI sector, where companies have broadly emphasized lower-cost, open-weight models. A DeepSeek researcher argued in a widely shared Sunday essay that advanced AI should remain broadly accessible rather than controlled by a small group of companies, and expressed distrust of Anthropic and OpenAI as potential gatekeepers.
Narrow Talks Remain Possible
Preliminary US-China negotiations were reportedly underway ahead of Chinese leader Xi Jinping’s planned Washington meeting with US President Donald Trump later that month. Areas of overlap include cybersecurity, model misuse and unexpected behavior from increasingly capable systems.
Researchers cited by WIRED said narrowly defined discussions could still produce incremental progress, but effective agreements would require credible ways for both countries to verify compliance. They also warned that rhetoric about weakening China could undermine the trust needed for those talks.
Source: China Isn’t Buying Silicon Valley’s Call for an AI Slowdown
SIGNAL TITANS
Leave a Reply