China Rejects Silicon Valley's AI Slowdown Proposal

China Sees a Bad Deal in AI Slowdown Talks

Silicon Valley's message on artificial intelligence has shifted dramatically. After years of warning about an AI arms race with China, prominent tech executives now argue the technology has become so dangerous that both nations must cooperate to slow development. But China isn't buying it.

Anthropic CEO Dario Amodei laid out this new position in a blog post published Saturday, writing that "global pacing will require cooperation with China, the autocratic country with by far the most advanced AI capabilities." The proposal quickly drew skepticism from Chinese government officials and AI researchers who see a fundamental contradiction in the approach.

The problem isn't that China dismisses AI risks or wants to race ahead unchecked. Rather, Beijing views the proposal as inherently unequal—a framework that would constrain Chinese capabilities while preserving American advantages.

Why it matters

The disconnect reveals a critical obstacle to international AI governance. If the world's two leading AI powers cannot establish basic trust around cooperation frameworks, efforts to manage existential risks or prevent misuse will struggle to gain traction. The impasse also highlights how national security concerns and competitive dynamics complicate technical safety discussions.

A Framework Built on Leverage

Amodei's proposal explicitly calls for the US and other democracies to maintain "as large as possible" a lead over China in AI development. He recommends cracking down on chip smuggling and preventing Chinese companies from using unauthorized model distillation techniques.

Only after weakening China's position, Amodei argues, would an agreement become "more likely in the future." He later acknowledged in a television interview that he has "no idea how to negotiate with China."

At a Monday press conference, Guo Jiakun, spokesperson for China's Ministry of Foreign Affairs, described the proposal as "fear-mongering, confrontation, and vicious competition" that would hamper global cooperation efforts. The ministry emphasized that "China puts equal emphasis on development and security in terms of AI."

Different Risk Priorities

Beijond the structural issues with Amodei's proposal, China and the US focus on fundamentally different AI risks. Silicon Valley's loudest voices emphasize existential threats—scenarios where rogue AI systems could pose catastrophic dangers to humanity.

Beijing's concerns center on domestic stability and Communist Party control. In a recent essay, China's minister of state security Chen Yixin warned about AI-fueled political disinformation, cyberattacks, espionage, and social unrest—spending far less time on doomsday scenarios.

Chinese AI companies, which have focused on releasing lower-cost, open-weight models, also view US safety rhetoric with suspicion. A DeepSeek researcher wrote Sunday that he believes calls for AI slowdowns serve to inflate company valuations or disadvantage competitors. "I do not trust Anthropic or OpenAI to do that," the engineer wrote regarding making advanced AI broadly available.

Room for Progress

Despite the tensions, cooperation isn't impossible. Preliminary negotiations are reportedly underway ahead of Chinese leader Xi Jinping's planned visit to Washington later this month to meet with President Donald Trump. Both governments share concerns about cybersecurity threats, model misuse, and unexpected AI system behavior.

What could move discussions forward, according to experts, is designing credible verification mechanisms and focusing on narrowly scoped issues where interests genuinely align. But rhetoric about weakening China may undermine the fragile trust such talks require.

These details were first reported by WIRED.