China Calls the AI Slowdown Plan a Cold War Tactic
3 min readDario Amodei asked the AI industry to ease off the accelerator. Beijing answered in under 48 hours, and the reply was not a counterproposal. It was an accusation.
What Amodei Actually Proposed
Anthropic chief executive Dario Amodei published an essay titled “We Must Pace the Frontier” on September 12. Its argument is that labs should slow the rate at which raw model capability improves so that safety research and risk controls have time to catch up.
The essay also carried a caveat that made the international reaction predictable. Amodei wrote that any slowdown would be bounded by the lead US labs hold over what he called authoritarian regimes, chiefly the Chinese Communist Party. Slow down by more than that margin, he argued, and unpaced Chinese projects pull ahead. Speaking to CNBC a day later, he called China the toughest dilemma in the entire plan.
Beijing Rejects the AI Slowdown Call
Foreign Ministry spokesman Guo Jiakun dismissed the framing on September 14, saying that fearmongering, confrontation and vicious competition would only disrupt global AI governance and serve nobody’s interests. State outlets went considerably further.
The Global Times described the proposal as a page from the Cold War playbook, aimed at restraining China’s technological rise through technical barriers and regulatory monopolies. A China Daily editorial made the commercial reading explicit, casting the slowdown campaign by US executives as a way to blunt competition rather than a genuine safety measure. The common thread across all three responses is the claim that a pause negotiated among American labs would simply lock an American lead in place.
Why It Matters
The exchange exposes the structural problem with any voluntary pacing agreement. A slowdown reduces risk only if the labs that matter all take part, yet the proposal was written with an explicit exemption for competitive pressure from China. That makes it easy to read as strategy wearing safety’s clothes, which is exactly how Beijing chose to read it.
Watch for whether the argument moves from editorial pages into policy. China runs its own frontier governance track and has signed international AI statements before. If Beijing formally declines to discuss pacing at all, the practical ceiling on any coordinated slowdown becomes whatever US and European labs can agree among themselves.
The safety case for pacing has not changed this week. What changed is that the geopolitics of it are now fully in the open, and the toughest dilemma Amodei named has answered back.