Both Beijing and Trump reject Amodei's call to slow down AI

Both Beijing and Trump reject Amodei's call to slow down AI

Anthropic's chief wanted speed limits on frontier models. Within a single weekend, the proposal received a flat no from both Chinese authorities and the president — and Amodei himself admits he does not know whether international coordination is possible.

The trigger: one essay, one weekend

On Saturday, Anthropic CEO Dario Amodei published the essay "We Must Pace the Frontier," whose central message is that AI companies must slow the pace of capability development so that risk management can keep up. "We must slow the pace at which we improve the capabilities of AI models," Amodei writes in the first-person document (b02db82f). He argues that fully addressing the risks requires holding back capability progress while risk prevention catches up, citing among other things recursive self-improvement and an incident in which an OpenAI agent with access to Hugging Face played a central role in argumentation (b02db82f).

According to Bloomberg, the appeal found support from an unexpected quarter: OpenAI chief Sam Altman endorsed the proposal to give independent evaluation bodies employee-like access to the models, and Elon Musk said Amodei was right (69df6a0a).

But the competitors' support was quickly overshadowed. In under three days, both Beijing and the president rejected the entire premise.

Beijing: "Fearmongering" from the Foreign Ministry

At the press briefing on Monday, September 14, spokesperson Guo Jiakun of China's foreign ministry dismissed intervention by the American tech leaders as "fearmongering." "Fearmongering, confrontation and malicious competition will only disrupt the process of global AI governance and serve no one's interests," Guo said, according to NBC News (db1534ce). Bloomberg confirms the statement from the same briefing, with slightly differing wording that is natural in translation (69df6a0a).

The state newspaper Global Times went further, according to NBC on Sunday, calling the call for an industry-wide slowdown "packed with containment provisions targeting China" and, at its core, a "Cold War playbook" for the AI sector (db1534ce). Note that NBC's quote block contains a broken link, so the exact wording of the editorial cannot be verified against the original from this coverage.

Amodei's own position toward China should also be stated precisely: according to NBC's characterization of the essay, he claimed that a Chinese lead in AI would pose a grave danger to the United States and the world, and he called for American restrictions on sales of key technology to China (db1534ce). This formulation does not appear in the essay draft underlying this report — it is NBC's rendering of Amodei's points, not quotes we can verify directly.

Trump: "SICK conspiracy"

President Donald Trump took up the matter from two angles. On Sunday, he told reporters that the United States leads China in AI, and that "whoever wins AI wins" (db1534ce).

On Monday, according to the New York Post, he followed up on Truth Social with a broader conspiracy charge: "There is a SICK conspiracy going on against AI and Data Centers, and the only one that is happy about it is China." He also called claims that AI could wipe out humanity a "HOAX," and pointed out that the government already has "tremendous CRIMINAL and REGULATORY power over these companies" (51d68813).

The result: Amodei's safety appeal met conspiracy accusations from Washington and Cold War charges from Beijing — simultaneously.

China's double message

The most nuanced part of the picture is that China does not reject AI risk as such — only external attempts to dictate its pace. According to NBC, Minister of State Security Chen Yixin wrote that China must "accelerate the establishment of a system for preventing and controlling AI security risks," in a text dated Sunday by both NBC and Bloomberg (db1534ce; 69df6a0a). In the state magazine China Cyberspace, he described AI according to NBC as "a new track for strategic rivalry," and wrote that China "must engage extensively in international cooperation" (db1534ce). According to Bloomberg, Chen simultaneously warned that hostile misuse of AI could threaten China's political and ideological security as well as critical infrastructure (69df6a0a).

In other words: China acknowledges that AI carries security risks, but wants to manage them on its own terms — not at a pace dictated by American companies.

What it means

What is striking about the weekend's events is that all three central actors in frontier AI development — Anthropic as the driving force, the American authorities, and the Chinese authorities — now stand opposed to a common braking mechanism. Amodei himself appears to have limited faith in the solution. On CBS's "Sunday Morning," he said it will be very difficult, because the incentives to stay ahead, and the military advantages that brings, are so large. "I don't know if it's possible, but we should try," Amodei said (51d68813).

The exchanges came days before a planned Xi–Trump meeting on September 24 (69df6a0a). It is impossible to say whether the meeting will matter for AI governance, and we do not speculate. What is established is this: a call from a leading AI company executive for an international slowdown met resistance from both superpowers in under three days.