China Called US AI Warnings Fearmongering. Then Its Own Spy Chief Said AI Could Topple the Party.

Key points

• Anthropic CEO Dario Amodei published an essay over the weekend arguing the US should actively impede China's AI advancement.
• China's government dismissed the essay as "fearmongering" and rejected any right Washington claims to block Beijing's AI sector.
• China's top spy chief separately warned that AI could threaten Communist Party rule.
• The two positions landed in the same news cycle, exposing a sharp internal contradiction in Chinese AI policy.

Dario Amodei, chief executive of Anthropic, the company behind the Claude family of AI assistants, published a lengthy essay over the weekend arguing Washington should do more than race ahead in AI. He wants the US to actively block China from keeping pace, while also calling for a global slowdown in how fast AI systems are made more capable.

Amodei has been the loudest major-company voice for safety-first AI development. Two stories we published on 12 September covered his push to give independent evaluators permanent access to Anthropic's models and his hope that the rest of the industry will follow, including his call for outside inspectors inside his own company. This latest essay extends that argument into geopolitics.

Beijing's official spokespeople called the argument "fearmongering" and rejected the idea that Washington has any right to impede China's AI sector. Standard diplomatic language. What happened next was not.

Why does the spy chief's warning matter so much?

China's top intelligence official warned that AI could threaten Communist Party rule. That is a striking admission from the same government pushing hard to lead in AI development.

The collision is the story. Beijing publicly dismissing foreign AI concerns while its own spy establishment privately flags the technology as a domestic threat reveals the bind Chinese leadership is actually in: slow AI down and risk falling behind the US; press forward and risk building something that is hard to control at home. No amount of diplomatic denial papers that over.

For people watching from outside both capitals, the practical consequence is a technology cold war that will shape which AI products reach which markets, how much they cost, and what safeguards they carry. Export controls, the rules governments use to restrict sending sensitive technology abroad, already limit which AI chips American companies can sell to Chinese buyers. Amodei's essay signals pressure to tighten those restrictions further.

What does this mean for people outside the US and China?

Everyone who uses AI tools feels this indirectly. If the US restricts the chips and software China needs to build advanced AI, Chinese developers will have fewer resources to release competitive products globally. That could narrow the field of AI assistants and productivity software available to consumers worldwide, or slow the pace at which prices fall.

There is no legislation on the table yet. Amodei's essay is a policy argument, not a law. Anthropic has lobbied in Washington before, though, and a CEO carrying this message publicly adds real weight to the conversation inside the US government right now.

The real news here is not what Amodei said. It's that China's own intelligence establishment is saying the quiet part loud. A government simultaneously dismissing foreign AI warnings and worrying privately that the same technology threatens its grip on power is not a government with a settled AI strategy. Watch what Beijing does with domestic AI regulation over the next six months. That will tell you far more than any diplomatic denial.

Common questions

What is Anthropic, and why does its CEO's opinion matter?

Anthropic is a US AI safety company that built the Claude line of AI assistants. It is one of a small number of companies shaping how governments think about AI risk, which means its CEO's public essays carry policy weight beyond a typical tech blog post.

What are export controls, and how do they affect AI?

Export controls are government rules that ban or limit sending specific technologies to other countries. In AI, they currently restrict the sale of advanced chips, the specialised processors needed to build powerful AI systems, to China. Broader controls could slow China's ability to build competitive AI at all.