Anthropic’s Dario Amodei answers: He’s not opposed to open-weight models, but he fears Chinese AI

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

“Anyone who has read my previous writings should know that I do not consider such a ban a useful measure, but let me state it clearly so that there is no doubt: Anthropic has never advocated a ban on open-weight models“, he wrote, emphasizing his words.

His response came after Nvidia founder and CEO Jensen Huang took to

While that letter did not specifically mention China, controversy in the industry centered on allegations that Chinese AI labs grow in capacity often by stealing intellectual property from their American counterparts. One method is distillation, where the AI ​​bombards another model with prompts to figure out how it works.

Amodei said in a blog post on Monday that he sees open-weight models as falling into a different bucket than the threat posed by China. “Models with open weights that don’t have critical capabilities are a public good: they don’t cost anything besides the computation to run them, and they provide value to companies, developers, and researchers.”

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

He fears that “authoritarian governments” will build models stronger than those in the United States to achieve “permanent military superiority” and/or use artificial intelligence to oppress their people. He said the Chinese Communist Party 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” and not just cybersecurity attacks. In his view, open-weight models are more dangerous in those scenarios because it is difficult to place guardrails on them or monitor their use. He also noted, citing a UK AI Security Institute report, that once open weights are released, they cannot be withdrawn.

This goes against what open source advocates say: access to robust open models that aren’t controlled by a single entity helps defenders protect themselves.

Amodei listed measures he believes will frustrate China, including restricting its access to hard chips (which has been a long-standing US policy) and calling for a crackdown on distillation. The United States has threatened to impose sanctions on China if it determines that there is intellectual property theft related to American models.

Interestingly, Amodei also said he supports growing efforts, some of which are led by the United States, to create a model organization for safety testing, especially if the entire world, including China, agrees to submit to it.

“I think this idea is actually close to consensus: I’m encouraged by the Trump administration’s move in this direction in recent months, and by recent industry proposals that would apply such a test to the most capable models regardless of their country of origin or whether they are open or closed (excluding less capable models, such as those in startups and academia, entirely),” Amodei wrote.

He added: “Note that for the test to be effective, it must be global, which means that even the Chinese Communist Party must be involved in it. I think this may indeed be possible… and limited cooperation on preventing biological weapons used by artificial intelligence may be possible, because it is in China’s interest as well.”