Anthropic’s Dario Amodei responds: doesn’t oppose open-weight models, but fears Chinese AI

Anthropic founder and CEO Dario Amodei responded Monday afternoon to industry mutterings that his company is somehow supporting the U.S. government’s efforts to ban Chinese open-weight models, or possibly open-weight models in general.

“Anyone who has read my previous writing should know that I do not consider such bans as a useful measure, but let me say it clearly so that there is no doubt: Anthropic has never advocated a ban on open weight models” he wrote, emphasis on his.

His response came after Nvidia founder and CEO Jensen Huang addressed X (his first message on the platform) to share an open letter on Friday in which Nvidia and a long list of other AI companies, including Hugging Face, Meta, Microsoft, Mistral and Nvidia, urge policymakers not to impose broad “premature restrictions” on open-weight AI models.

Although that letter did not specifically mention China, the industry debate has focused on allegations that Chinese AI labs are increasing in capacity, often by stealing intellectual property from their American counterparts. One method is distillation, where an AI bombards another model with clues to learn how it works.

Amodei said in the blog post published Monday that he believes open-weight models are in a different category than the threat posed by China. “Open-weights models that do not have dangerous capabilities are a public good: they cost nothing other than the computing power needed to run them, and they provide value to companies, developers and researchers.”

Amodei said in his post that he has long had concerns about AI, but that companies using open-weight models, even those from China, are not among them.

His fear is that “authoritarian governments” will build models more powerful than those in the US to achieve “permanent military superiority” and/or use AI to oppress their own people. He said the Chinese Communist Party (CCP) is not the only authoritarian government he is concerned about, but it is the “most capable.”

Amodei also said he fears AI will enable “biological attacks,” not just cyber attacks. And he says open-weight models are more dangerous in those scenarios because it’s difficult to put guardrails on them or monitor their use. He also noted, citing a report from the UK AI Security Institute, that once the open weights are released, they cannot be withdrawn

This goes against what open source advocates claim: that access to powerful open models that are not controlled by a single entity helps defenders protect themselves.

Amodei listed actions he said would thwart China, including restricting access to high-performance chips (which has long been a US policy) and calling for formal action on distillation. The US has threatened sanctions against China if it detects IP theft involving US models.

Interestingly, Amodei also said he supports this growing efforts, some led by the US, to create a model safety testing organization, especially if the entire world, including China, agreed to submit to it.

“I think this idea is actually close to a consensus: I am heartened by the fact that the Trump administration has moved in this direction in recent months, and by recent industry proposals that would apply such tests to the most capable models, regardless of their country of origin and whether they are open or closed (while completely excluding less capable models, such as those from startups and academia),” Amodei wrote.

“Note that to be effective,” he added, “testing would have to be global, meaning even the CCP would have to participate. I think this could actually be possible… limited cooperation around preventing biological AI weapons could be possible, as this is also in China’s interest.”