Anthropic CEO Dario Amodei called for a global slowdown in AI development, arguing that the US and other democracies must maintain a lead over autocracies like China. He emphasized the need for cooperation to prevent AI from becoming too dangerous.
The proposal has drawn sharp criticism from Chinese officials and AI researchers, who see it as a veiled attempt to gain leverage.
Amodei's blog post suggested that slowing AI progress could increase the US's advantage over China, which he believes is the only way to ensure global safety. However, Chinese officials and experts argue that the proposal feels inauthentic, as it implies China should be restricted before being included in discussions.
Kevin Xu, founder of Interconnected Capital, noted that the Chinese AI community shares many safety concerns with US labs but resents being treated as a competitor rather than an equal. "The gulf lies between an Anthropic-led vision that would only subjugate and restrict China first to maximize leverage before talking," he said.
"We Must Pace the Frontier," Amodei's essay argued, framing the slowdown as a condition for maintaining a lead over China. However, the proposal has been criticized for its lack of clarity on how it would encourage Chinese participation. Amodei later admitted he has "no idea how to negotiate with China" during a television interview.
Chinese officials, including Foreign Ministry spokesperson Guo Jiakun, rejected the proposal as "fear-mongering, confrontation, and vicious competition." They argued that the call for AI safety ignores China's focus on maintaining social stability and the CCP's control. "China puts equal emphasis on development and security in terms of AI," the ministry stated.
Michelle Nie, a visiting fellow at the Center for a New American Security, noted that Chinese policymakers are skeptical of the idea that AI poses existential risks. Instead, they are more concerned about AI threatening the CCP's rule.
"Beijing finds it difficult to take these calls at face value partly because it’s skeptical of the possibility of achieving AGI," she said.
Source: wired
