China state media calls Amodei’s AI slowdown essay a ‘Cold War playbook,’ Reuters reports

Reuters reported that China’s state-backed Global Times said Anthropic CEO Dario Amodei’s essay calling for slower frontier AI development is really a “Cold War playbook” aimed at China — calling the push “hypocritical and short-sighted” and arguing it tries to curb Chinese AI through tech barriers and regulatory monopolies.

Beijing’s state-backed paper answers the Amodei/Altman/Musk weekend by casting safety pacing as China containment — framing U.S.–China AI talks and any Trump–Xi AI agenda as contested politics, not shared slowdown consensus.

Reuters’ Laurie Chen, writing from Beijing on 14 Sep 2026, reported that China’s state-backed Global Times said Anthropic CEO Dario Amodei’s Saturday essay may look like a rational safety statement but is really a “Cold War playbook” targeting China. Global Times is a Communist Party–backed tabloid. That Reuters wrap is the filing event. This is an editorial attack as reported.

Reuters quotes the paper on what it called the true agenda: “to attempt to curb China's AI development through technological barriers and regulatory monopolies, uphold Washington's monopolistic hegemony in cutting-edge technology, and exclude China from the global AI governance system.”

Reuters also quotes: “This 'silent AI Cold War' is hypocritical and short-sighted.” The paper said excluding China would “significantly increase the trial-and-error costs and risks of loss of control in global AI development.” Loss of control here is the paper’s claim that cutting China out would make global AI riskier — not a new accident report.

Reuters, citing sources, says the United States and China are planning to discuss frontier AI safety risks in mid-September, and that the issue could come up at a Trump–Xi summit on September 24.

The Global Times editorial itself lists three containment steps it says Amodei wants first: a strict ban on selling powerful AI chips or chip-making equipment to China; a crackdown on so-called model distillation; and tighter security at AI companies to stop theft of model weights — the learned numbers inside a trained model.